I'd say that instead of making self-driving cars as reckless as typical drivers, we should keep the cars driving by the book, and fine the people who collide into them. Will it cause frustration? Yes, but even today, people who try to drive safely and with respect to the law cause frustration too. And I think self-driving tech has enough momentum now that, frustrating or not, they will get accepted anyway.
I agree. Maybe it is time that all vehicles (including cyclists) were required to carry dashcams etc and a mandatory report (that is as automated/cheap to generate as possible) required for every accident, no matter how minor.
That's an even worse idea. You've shifted all of the data from the agency at least tangentially concerned with people's welfare to a corporation only concerned with profit.
I don't think the cameras have to be network connected. Have the video feed locally only and for insurance claims you have to provide the relevant bit of video that will help determine the culprit. Like the Russians do.
That was not the point. Why do you think the Russians do this? Because the insurance co would deny any coverage by default ("meh, must have been your fault anyway, now f* off"), and no recourse.
The other way to look at it is that the traffic laws, taken to the letter, are too conservative in some circumstances, and there are perfectly safe, more efficient limitations that can be employed (The 100% complete stop for example, frivolous waste of potential energy!) Maybe instead of teaching self-driving cars to break the laws like everyone else does, we should change the laws to reflect more how people drive.
> there are perfectly safe, more efficient limitations that can be employed (The 100% complete stop for example, frivolous waste of potential energy!)
Genuine question: what do you think is the better alternative to the 100% complete stop? I'm not saying I think there is or isn't a better alternative, just wanted to know what you had in mind.
what do you think is the better alternative to the 100% complete stop?
I think that would be to slow enough that you can stop if you need to, but not actually stop completely --- something which almost everyone already seems to do, because it saves wear and helps with passenger comfort too. (I've been in vehicles where the driver insisted on completely stopping exactly as legally specified, with the accompanying double-jerk of deceleration followed by acceleration. Not pleasant at all.)
If the stop causes a "jerk", the driver is braking far too aggressively for an expected and routine stop (which implies that they'll have much smaller margin of safety when emergency braking would be called for). This seems to strengthen the point "people are driving like crazy, that's not a good thing".
Next time you drive, come to a complete stop and feel the car as it rocks back on its suspension. This will happen even if you brake quite lightly. You can slowly raise the brake so that you are barely braking right as you stop, which will almost completely eliminate the feeling, but it makes your stop excruciatingly slow. Theoretically, you could also eliminate the rock back at a stop sign by lifting your foot off the brake right as the car stops (before it moves backwards that little bit). The vast majority of drivers who think they are stopping completely actually lift their foot off right before that moment, so they don't experience the jerk.
> lifting your foot off the brake right as the car stops
You can do that even if you're braking very hard, and it doesn't have to make your stop slow. All you want to avoid is that bounce back right at the end, and backing off the pressure a bit just before the full stop will be enough to achieve that and create a very smooth ride. Of course you will always have the deceleration but for passengers that's far less jarring than the bounce that happens when the front suspension unloads when the car has stopped.
Fortunately it's easy to practice as stop signs are everywhere.
Exactly, it took a while to notice this was even possible. It does take a few more inches to stop. But, now I can't help but notice that some people get a lot more jerk
It's sort of an art, what with variable conditions and whatnot. Surprisingly, automatic trains manage to get it right, being programmed to ease on the brakes just right to avoid a jostle altogether. Fewer variables to keep track of, I suppose.
Yes exactly, much less going on. It's probably easier with an automatic system than a human brain too. A fairly simple feedback loop between an accelerometer and something to modulate brake pressure should create a pretty smooth stop without much difficulty.
There are already regulations allowing for that, but they are not used everywhere because some people abuse the privilege of being allowed to use their own judgement. If (and only if) everyone practiced good judgement and restraint, the regulations would not need to be so prescriptive (Full disclosure: I don't always practice what I preach.)
In the UK we have very few stop signs. In fact, in my entire life, I only recall ever coming across 2. We usually have a yield action at intersections so that you only even need to slow if the line of sight at the intersection is impeded to the point you cannot see the flow of traffic. Consequently there is no need to come to a full and complete stop unless the flow of traffic prevents your ability to turn where the yield sign actually requires that you come to a stop while you wait to merge into traffic. This functions far better than the ridiculousness that is stop signs that require you to come to a complete stop even if you are the only car on the road.
Nothing is more frustrating than coming to an intersection with full line of sight for miles in both directions, where you know there is no car that is close enough to hit you, and having to come to a full stop anyway.[1]
In my part of the US we have a lot of rotaries (round abouts). Thousands of cars go through them every day, no stopping and no accidents... it can be done.
[1] This is of course hyperbole. There are much more frustrating things.
Edit: In fact I thought of something better. Waiting 10 minutes at a red light when there is no one else on the road.
In europe most of intersections that would be all-way stop in US either do not have any signage or signage to remind drivers of the rules of non-prioritized intersection ("warning, yield to the right"). Small roundabouts were to large extent specific to UK, until various european local authorities found out that building roundabouts is good way to meet their EU grant quotas.
I’m a British person who had immigrated to North America. A four (or more) way stop sign is not replacing a yield at an intersection, it is replacing a roundabout. The reason you must come to a full stop is to signal to other drivers the time you arrived, since that is strictly the order that people use to go. It’s also a safety issue since you need to check right and left for out of control cars, bikes and so on. As s system I prefer roundabouts as they keep the traffic flowing, but anecdotally the stop sign seems much safer. Finally it’s a an enforcement issue. Cops don’t have to argue that you didn’t slow enough. If you didn’t stop completely you broke the law.
As a UK driver, generally all the problems seem to be solved by the "Give Way" signs (Yield) rules and systems; Major/minor road priorities for asymmetric traffic loads, T-junctions, and mini-roundabouts for equal loads but small volumes (large volume is handled by traffic lights or normal roundabouts).
Four-way stop signs seem to be a desperate attempt to either avoid telling American drivers to "not be selfish" and allow other traffic through ahead of themselves, or to just avoid having to have drivers learn priority systems.
Avoiding major/minor road intersections seemed to be the main use of four-ways (around West-Chicago/Illinois, where I did most of my US driving), with ridiculous cases like four-way stop signs on roads where you could literally see hundreds of feet down each of the other roads, and not a car in sight except the cop car parked next to the intersection.
True stop signs in the UK are reasonably uncommon except in cases where they are actually required e.g. a junction could appear misleading.
> with ridiculous cases like four-way stop signs on roads where you could literally see hundreds of feet down each of the other roads, and not a car in sight except the cop car parked next to the intersection.
Keep in mind that rotaries are extremely rare in the US, and that many four-way stops are replacing what would be rotaries in the UK.
Also, some intersections are more dangerous than they look. I remember a 4-way intersection with a 3-way stop that looked safe, but after the 10th traffic accident or so, the town said, "Let's just make it a 4-way stop," and the accidents stopped. One of entering roads was right after a major speed change (so people wouldn't slow down enough), and another had slightly worse visibility than it seemed. But once everybody was made to stop, the intersection worked safely, at the cost of adding 30 seconds of travel time.
This varies per implementation of roundabout. Plenty of the roundabouts I've driven through have stop signs and only allow one car in the roundabout at a time.
Others have been made in places (like a crevasse) where the circle is too small for a city bus to safely traverse - it has to jump the curb every time. The traffic light it replaced was actually slightly safer.
I guess if you compare a bad roundabout to an ideal four-way intersection that's true, but in general we can safely say roundabouts have much lower rates of accidents, especially serious ones, than do traditional intersections, despite many drivers' intuition that it should be the other way around.
I would say that if it has stop signs it's not a roundabout -- the concept requires that incoming traffic yields to traffic on the roundabout but need not stop.
Indeed. With 100% robots on the road, we can relax the rules to what machines can safely handle. Which will mean much denser and faster traffic, because a) machines can handle more than humans, and b) machines don't (unless programmed to do so) routinely disrespect the rules like humans.
> Maybe instead of teaching self-driving cars to break the laws like everyone else does, we should change the laws to reflect more how people drive.
In my experience, people always drive a bit faster than allowed. If you raise the limit, people raise their speed as well. Maybe the solution is to only allow the relaxed rules to robotic drivers? (though that would also raise resentment).
>In my experience, people always drive faster than allowed
I think its more of people prefer to drive at the speed they're comfortable with (they believe its sufficiently safe given the environment), and legal speed limits tend to be significantly lower than that, except during special cases ie traffic jams. Im not sure if people are actually good at guaging where they should feel comfortable driving, but if you relax the rules for AI, it would be in a similar setup and the same question would apply.
With AI you can determine the accuracy of its "sense of comfort" through design and testing; it's harder with humans.
But the thing is, that on the road people often drive not at the speed they're personally comfortable with, but closer to the average speed all drivers on a given road stretch are comfortable with. I.e. I've often been in cases where the driver drove faster than he/she liked because otherwise everyone else would stick to their tail, honk, or aggressively pass by.
I think you're thinking about it the wrong way. Following the letter of the law doesn't mean the ride is safer. For example, say the software could say for certain that there is nobody else in an intersection. Thinking about how you would design a new system of traffic rules from scratch, why stop for the stop sign?
... because there is a stop sign? you should stop there nevertheless.
major number of motorcycle accidents happen because people feel they have overview of whole situation and don't respect the simple traffic rules, both in cars and on bikes.
and yes, based on my 100k+ ridden mostly on european roads, following dead simple traffic rules of these days would prevent many accidents (and deaths). Maybe we would all get to destination 3 minutes late... so what (actually self driving should clean up the roads and especially traffic jams)
"It's completely safe, there's nobody else at the interse-CRASH Sorry mate, did not see you." You're impressively sure in your assumption that software can be certain that the intersection is empty.
If the car can't be certain that the intersection is empty, it's going to hit something when it tries to cross regardless. With fully networked cars, sensors in the intersection, and tons of sensors built into the car, the software can be pretty damn sure that the intersection is empty... at least more so than a human can.
Perhaps. Or you now have to deal with actual obstacles, and with obstacles that are denied or falsely inserted by rogue nodes. "Fully networked" is a mixed bag.
When there are no human drivers anymore, we can relax the rules on the assumption that self-driving cars will communicate in realtime. Until that is, we need to deal with the fact that a human driver can not say for certain an intersection is empty.
A lot of the traffic rules exist to remove the need for having total overview of the situation. That is, if you actually follow them, and can safely assume everyone else follows them too, then you don't need to see what's behind the corner and you can be confident you'll get to your destination safely.
While that is a fair point I think a reasonable counterargument would be that the current rules are based around the fallibility of humans (stopping distance, speed limits, etc) which is less applicable to AI due to it's lower reaction times and greater awareness of the environment around it (in terms of it's constant sensor sampling when humans would normally "zone out").
Up until human drivers get removed from the public roads entirely, we need to apply human-centric rules, because humans are the weakest link in the chain. One of the problem with human drivers is that they feel they know better, and routinely disrespect the rules. Peer pressure makes it hard for individual drivers to follow the law when so many people get angry at them for that. Introduction of self-driving cars provides us with good opportunity to improve the safety of human traffic - autonomous cars don't care if you're angry at them, and by following the rules, they can force everyone else to do the same. That is, unless we chose to have self-driving cars ignore the rules too, like the article suggests.
Drivers are at least protected by their metal shell. What of the bikers, cyclists, and pedestrians? Those are unlikely to become non-self-driving, so to speak.
True. For that reason, I doubt self-driving cars will be going 200km/h through the city centre.
I can imagine autonomous motorcycles. The hardware doesn't take that much extra space to make it infeasible. Whether anyone would like to use such a vehicle is another matter :).
I'm sure there are some drivers affected by peer pressure but I would be surprised if it's a significant figure as there are already a significant number of drivers who do stick to the speed limits. You tend to see worse driving in areas where people feel safe (ie places they drive frequently) or when they are themselves in a hurry (eg "rush hour") but those are more personal to the individual rather than pressures placed on them by other drivers.
Furthermore, I'd agree that increasing the limits above what humans currently already do would be a bad idea. At least as long as it is still legal to have human operated vehicles on the roads. However the issue we are discussing is whether you want AI to emulate some of the more impatient behaviours of humans. In many instances I think it would be safe. So it seems logical to bring AI up to our level of throughput.
Eventually the long term goal of driverless cars is reduced congestion due to the improved throughput provided by AI. But so long as there are humans also driving cars on the road the only way AI could improve throughput is to emulate human behaviour to some extent (ie not the worst of human behaviour but more our ability to judge when we need to navigate tighter spaces in congested junctions - for example).
- Assume that drivers who engage in the "more impatient behaviours" are increasing the amount of accidents compared to what would happen if they stick to the rules.
- Note that introduction of self-driving cars is a perfect opportunity to use them as a forcing function on the road, to make those impatient drivers stick to the rules.
- If both of the above is true, we can save more lives right now by having the self-driving cars stick to the rules than by having them emulate the impatient drivers - a task which the machines will handle fine, but which would keep such behaviour legitimized for humans.
Now I'm least sure about the first assumption; it intuitively seems true, but beyond occasional articles pointing out speeding and careless driving as primary sources of traffic accidents, I haven't spend time evaluating the data on that.
The second point seems intuitively true from experience of how drivers adapt to other drivers around them. Self-driving cars can be made stubborn by default, so everyone else would have no choice but to adapt to them.
I understand your point but I don't agree with your conclusion. You cannot assume that human drivers will drive any more differently when there are AI cars on the road sticking to the limit than they already currently do with non-AI cars on the road sticking to the limit. Either way they're still going to want to drive impatiently.
Take learner drivers for example, people sometimes drive worse around them (pulling out in front of them etc) because they simply don't want to get caught behind them. This is even before they've sampled the competency of the learner. The L plate is often enough to make bad drives drive worse.
Thus the only definite way you can argue that AI would slow down bad drivers would be if the bad drivers were the ones being driven by the AI. In which case we are back to the point about whether AI can safely drive a little more aggressively (for want a better description) due to it's greater competency at driving.
"They drive funny" claims human, regarding high level of reporting of accidents where computer drivers are required to report encounters but humans are not.
If humans were required to report every bump and scrape, what would we find out? I suspect we would find that humans are just poor drivers overall, and "they don't drive like I do" is the number one cause of minor incidents worldwide.
My main submission to this argument is the ubiquitous rear-ending in the turning lane where the first car's driver starts to roll and then stops, while the second driver is turned to watch approaching traffic and thus oblivious to the behaviour of the car in front.
"They aren't driving like I would" is a major red flag for me, causing me to either drive/bike/walk more carefully, get well away if possible, and keep them in view if necessary.
The issue isn't that the cars are worse drivers than humans. They're obviously going to be much better statistically. The article is just making the point that they're going to have to be programmed to break the rules just like humans do i.e. with speeding, stopping too often etc. If they are not programmed to break the rules, they will frustrate people and even cause accidents as humans try to interact with them. There's not always a good way to know that the car in front of you is being driven by a computer.
This will be an opportunity to fix the traffic rules to be more realistic. Often, the speed limits are too low in the US, and "everybody" goes faster. This leads to the problem of selective enforcement, and people not trusting speed limits when they are really neccessary.
In contrast, I've found most speed limits in Germany to be reasonable absolute maximum limits - except on one or two places on streets I know very well. And I'm not talking about the Autobahn - if I see 70 km/h in Germany (45-ish mph), its probably dangerous to go faster. If I see 45 mph in the States, it seems to mean "everybody go 55 mph".
1. Do you want to be the local politician who says "hey, let's raise this to 55 MPH" and then on a bad weather day or just out of sheer bad luck soon after there's a major pile up and then said politician gets roasted?
1. The traffic lights need to be coordinated and that means presuming a certain speed. Now the problem is a lot of traffic lights are close enough that changing the programming of just one should ripple through the entire system and this recently (especially on a regulation making time scale) was a problem consuming extraordinary amounts of computing power.
Speed limits aren't just about safety. Lower speeds mean less noise, which is desirable to local residents in residential areas. They're the ones who local politicians are accountable to.
Decreasing speed limits is something that local politicians can use to demonstrate to their constituents that they care about them, but very rarely has net-positive effect on noise and polution.
Pollution too. Speed limits are sometimes lowered when it is a problem. Temporary, during peaks, or permanently.
The other, more cynical reason for lowering speed limits is that in many countries now, speed traps are a significant source of income for the state. Making the speed limit significantly lower than what the standard 85% rule says is a good way to extract money from unsuspecting drivers.
but this is exactly the sort of political cover the politicians need. nobody wants to be the politician that says "lets raise the speed limit here to 55MPH", but if everybody drives 55 and all of a sudden there's a bunch of self-driving cars going 50 and people are getting annoyed at the road getting blocked up, changing traffic rules to conform to reality becomes a lot more possible.
> Often, the speed limits are too low in the US, and "everybody" goes faster.
Excessive speed is the primary cause of a third of all fatal accidents and a contributing factor in at least half. Instead of training the self-driving cars to be more human, we ought to train humans to act more like the computers! Speeding is one of the stupidest norms that exists in our society. We should definitely not try to encourage it.
Yeah, that's not a great study (though it was admittedly pretty groundbreaking,) and it doesn't actually prove that there is no relation between speed and accidents. Multiple studies since the 60's have shown different results.
>Politicians arbitrarily setting the speed limits is the root cause.
> Yeah, that's not a great study (though it was admittedly pretty groundbreaking,) and it doesn't actually prove that there is no relation between speed and accidents. Multiple studies since the 60's have shown different results.
The link sourced multiple studies. The originals were dubious, subsequent ones not so much.
But the issue the Solomon curve and other similar studies points out is that humans (and largely Americans) will ignore the rules if they feel they can safely go faster.
And sure, you can't ignore overall average speed. Crashes on roads where the average speed is 100 mph are going to be much more prone to fatalities than 55 mph roads. So, there are two rough equations: Lower average speeds mean less severe crashes; speed limits set to the 85th percentile mean fewer crashes. As the speed limit is raised (from 0), there will be at least one solution that satisfies both of those equations and results in the fewest fatal crashes (if that's your goal).
In certain situations the lower speed limits can make things less safe. For example, a straight, well maintained toll road with wide lanes, clear vision ahead, and say wildlife barriers that was designed for 75mph max speeds is constructed in a state with statewide 55mph speed limits. Construction uses the latest traffic easing and safety techniques, since you don't want to construct all roads to enforce speed limits that may change in the future.
So now most truckers and cautious drivers go 55mph, while people with less inhibitions are weaving in and out of them.
Are you talking about the wikipedia link? All the subsequent studies save one that it sourced disproved the first's premise or drew attention to serious questions as to its validity. Did you read them?
Yes, but the speeding norm is partly caused by speed limits being uniformly too low in some circumstances leading to speeding being tolerated to a point and everyone making their own judgements about how fast is too fast. It would be better to have higher limits that are strictly enforced rather than lower limits that no one follows.
Did you bother reading what he said? Speed limits are lower than what is safe on many roads. Limits should be increased to the safe maximum and then the maximum should be more strictly enforced. Minimums should be enforced as well, since driving significantly below the maximum is also unsafe.
It doesn't mean increasing limits to something that is unsafe.
Counterpoint: as a German, I know that when a sign says 50kmh, everyone goes 60, and when it says 70, everyone goes 80, etc. If you go as fast as the sign says, you will most likely cause traffic.
> If you go as fast as the sign says, you will most likely cause traffic.
Counterpoint: creating traffic by going at the speed limit requires that other people on the road are going over. Both parties are to blame, but only one party isn't increasing their energy at a exponential rate (1/2mv^2).
Your interpretation of American speed limits is quite correct. Driving the speed limit on a major highway in any but the rightmost lane is dangerous due to the speed other drivers expect you to be going.
On the flip side (almost literally), I once drove around Ireland for a week. Of course my instinct was to drive slightly over the speed limit, but I quickly found 100 kph on 1.5-lane roads winding around sheep pastures with no visibility nearly impossible to safely match let alone exceed…
They should follow the law. Granting an exception is not the way the law should work, because such exceptions are arbitrary and without any sense of justice or ethics. The best way to fix a bad law is to consistently enforce it.
If two actors, one of which is following the law, has a higher risk than two actors both breaking the law, the lawbreaker is still at fault for the increased risk.
That, or... if enough AI drivers are out there then maybe humans will just learn to live with them.
While crashing into an AI driver I may be able to blame the AI, it's still a huge pain and possible loss for me as a human.
If the reality is that AI is not leaving and they have peccadilloes, then the human drivers and legal structure will just need to adapt (something we aren't bad at).
Computers have been training humans for the past couple decades, and I see no reason for that to stop just because the computer's form factor is a car.
If you take a system where everybody does similar and add something that does not, this is going to cause problems. This is going to cause problems even if the dissimilar behavior is the proper legal behavior.
Erratic behavior causes accidents, be it driving too slow, too fast, weaving, impaired, etc...
Now, if all the cars drove the same way, we'd be fine. We'd have very few accidents. However, we don't. So, we bump into stuff.
From the point of view of drivers who're used to everybody bending the rules to a certain level, strictly following the rules is at the very least unexpected; 'erratic' is perhaps an overly strong word but I think the general point is sound.
If it helps, you can call it things like non-standard.
We don't actually tend to follow all the rules of the road. Like all changes in traffic, we will adapt. It takes time. Something like a rotary can take 18 months before traffic fully adapts. I can't speculate as to how long this will take.
I've been in 3 minor car accidents, ALL 3 have come where someone read-ended me in this way. I am constantly conscious of this and am always keeping an eye on the car behind me when I'm in this situation.
Yes, it's relatively simple things like this that can actually improve car safety right now, no self-driving capabilities necessary. Automatic braking, lane departure warning, dead-spot detection, driver attention detection, missed signal warnings, etc.
I don't understand why we are not investing all of the efforts and resources that are now spent to develop full self-driving capibilities, into making existing viable safety technology cheaper, and eventually mandatory. I'm 100% convinced it would have a much bigger impact on traffic safety, in a much shorter timeframe, to the point we don't even need self-driving cars anymore. At least not for safety purposes.
Come to think of it, maybe the whole self-driving car hype is not actually because companies want to improve car safety, but because they see new business models based around them, and we, the public, need to be convinced self-driving cars will be safer than alternative technologies or modes of transportation.
It's certainly because of the potential market. Nobody is even claiming to be researching self-driving cars for the safety, it's always because of the potential.
Yes, AEB and Adaptive Cruise are the two most important features I looked for in a car. I ended up leasing an EV without them (waiting for my Model3), but my next vehicle will require these (for example, almost all Toyota models have similar features - TSS-P warns for frontal crash but doesn't brake immediately).
Since beginning of this year I almost exclusively drive cars with some kind of front facing radar for AEB/ACC. On one hand there were situations where imminent collision warnings or even AEB were useful, on the other the amount of false positives for collision warning is annoying and in some cases actually dangerously distracting.
And for some reason the minimal front distance for WV's ACC is almost an order of magnitude larger than safe braking distance, so other drivers see no issue with changing lanes into that large free space in front of you.
So in all, this technology obviously requires some work both in terms of reliability and in terms of UX.
Come to think of it, I've been hit from behind this way twice, and done it to someone else once.
I'm sure self-driving behavior is a factor here, but I also suspect that breaking and passing at intersections is just a common cause of zero-damage accidents. After all, one of the collisions in the story was a bike rear-ending a car - how many drivers would report that if there was no damage?
I've heard (citation needed) that if you are going to be in an accident the best step to take is to relax.
Anecdotally, it depends on how far away you see it coming. I'm constantly scanning front, rear mirrors, and side mirrors and if someone is driving erratically I take steps to get out of their way (in a safe and predictable manner of course!) I have no evidence this keeps me safer, but this and other "traffic calming" behavior I engage in makes my commute less stressful. It has the nice side-effect of continuous practice that reinforces good driving techniques.
"if someone is going to rear-end you, there's not much you can do (even if you can see it coming), other than brace yourself"
.. and by "brace yourself", you would touch your head to your headrest so that your neck has nowhere to move/flex with the impact. You want all of your body to be completely rearward in your seat.
Also, while this may be morally negative, if there is another car in front of you, you could creep forward and lightly touch their car with your own, making yourself the middle ball in the Newtons Cradle[1] that your three cars now describe.
Unless you can guarantee contact, is moving towards another car a good idea? Intuitively (which of course could be _completely_ wrong), it doesn't seem like being forced to decelerate sooner would be useful.
1. Why don't people pay attention to break lights? It's hard for me to relate to people who are always stopping last minute.
2. When someone is driving aggressively or erratically behind me, I break-check them as often as a I feel I need. Won't solve all rear-end issues, but it improves your chances of avoiding accidents.
One of the most common accident modes is right-turn-on-red. The first car will stop (or not), creep forward, etc. If they see an opening -- or think they seen an opening -- they'll let off the brakes and begin to roll under idle power or even accelerate slightly. Of course, this is the cue that they're making a right on red, so the 2nd car in line will also accelerate and at the same time look to the left for oncoming traffic. And BAM, you rearend the car in front of you because you only thought it was pulling out. It's the perfect storm of misreading the situation, and the situation demanding that you look somewhere other than where you're going.
> If humans were required to report every bump and scrape, what would we find out?
This seems like an awfully good point. One of the incidents described was a bicycle rear-ending a car - reporting from that incident reveals a scraped knee, a scratched fender, and an intact bike. How many times a day do equally minor accidents between people get ignored?
Maybe self-driving cars really do get rear-ended more, but I haven't seen anyone try to account for reporting gaps like that.
I wrote something similar on this subject a few days ago.[1] Autonomous vehicles (the good ones) have sensor-generated maps of what's around them, and those maps indicate what's unknown due to occlusion. Cautious behavior at intersections comes from that. If visibility is good, there's no hesitation.
This is fine. There are no CA DMV reports of "autonomous vehicle entered blind intersection and was hit by cross traffic". Near term solution: put a rear bumper on your autonomous vehicle which meets the old 5MPH bumper standard from the 1980s. Longer term solution: manually driven cars need basic anti-collision radar, which is now standard on many high-end cars and all Volvos, and may be mandated in the US just as ABS is now.
Useful temporary fix: brake lights that go up to super-bright ambulance level when the autonomous vehicle's sensors indicate something close behind and there's a need to brake.
"Useful temporary fix: brake lights that go up to super-bright ambulance level when the autonomous vehicle's sensors indicate something close behind and there's a need to brake."
this seems like a pretty clever solution if it would actually encourage the correct behaviour in human drivers. unexpectedly bright tail-lights might just have the opposite effect though, and further distract the driver who is about to rear end you.
When I was in university we made a project for the Statistics class: Measure the velocity of cars from bridges for cars going outside and inside a big city in Europe.
We controlled all big highways.
The result: Over 95% of the cars were going way faster than what the rules were and nothing wrong happens. That is, rules are wrong because politicians decide them and they are always extra cautious, so they are not responsible for accidents, and they make money out of fines.
I see a big problem if everybody follows the law every time, like robots do, because rules are so wrong in the first place.
One of the big things with robots is that you could have real feedback with real data measuring in the real world. You don't need to guess, just test in the real world and measure the consequences for example, of increasing the velocity.
You could also change the parameters on real time when you know there is ice or snow in the road... you don't need to have the same parameters when it is sunny in summer and in winter.
The result: Over 95% of the cars were going way faster than what the rules were and nothing wrong happens.
I'm not sure what do you mean, there are plenty of accidents caused by speeding both on highways and within city limits. Care to elaborate how you conducted the study and/or link to it?
As far as I'm aware extra caution is built in exactly to account for human errors and disregard for laws and have a margin of safety.
I can't help but feel maybe your professor missed an opportunity to teach about sampling errors.
A road can go decades without a crash, reduced speed will reduce severity of a crash when it happens.
Reduced speed usually increases fuel efficiency, which is also a factor in pollution control; in cities in particular that can have a big impact on health.
You assess the road conditions and conclude that the speed limit is set too low for a road that is in broad daylight and has no other cars. Are you implying that the speed limit is binarily safe, no matter the road conditions (night, rain, snow, congestion, etc)?
I doubt many drivers know their car and tyres well enough to assess the safe speed given current conditions. People either adapt to others or choose a speed they feel comfortable with. Speed limits are not perfect for every condition but humans are certainly not good at choosing a better one while driving.
This might be a problem with the ways roads are "signed". I believe that stop signs are much more common in the US than in Europe where there are signs that are more relaxed. I.e that you have to stop only in case of another car.
> "Humans violate the rules in a safe and principled way"
I don't share this optimism. I've been driven by folks who violate the rules in totally unsafe and scary ways (though "principled" in a way, in that they do it consistently -- unless a cop is looking).
Spend some time driving in Latin America or West Africa.
The rules mean nothing, everyone drives freestyle. In towns and cities, it works extremely well because everyone is very attentive and ready for anyone else to do anything, at any time.
My experience in over 60,000 miles in those parts of the world through more than 30 countries would say they are an order of magnitude less crashes there than in US/Australia/Europe, in the towns and cities.
(Note: highway driving is a whole different story)
> The rules mean nothing, everyone drives freestyle. In towns and cities, it works extremely well because everyone is very attentive and ready for anyone else to do anything, at any time.
There's some truth to that, but there's also the ugly reality: statistically those places tend to have extremely high death rates (which may or may not come with high accident rates). Because while people are definitely more aware and cautious on the road in general, nobody will ever take the extremely dangerous drivers off the road, and they're now emboldened by the lack of consequences (both from the law and from drivers who mostly willingly give up their right of way to whomever is most reckless).
I know someone whose entire family was killed by a driver who hit them head on due to blindly passing a truck while going 160kph on a 40kph road in Honduras. I know for a fact that his wife was an extremely cautious driver, but you can't avoid something like that no matter how hard you try to predict the unpredictable.
> My experience in over 60,000 miles in those parts of the world through more than 30 countries would say they are an order of magnitude less crashes there than in US/Australia/Europe, in the towns and cities.
Do you have any statistics that indicate that there are 90% fewer accidents in Latin American and West African cities than in Western, adjusted for the number of drivers? Seems rather optimistic.
This article is a good example of stupid tech reporting. Someone who's supposed to be a knowledgeable reporter does some speculating, without bothering with any actual data, and comes up with a so-called "problem" and idiotic suggestion for "solution".
The article suggests that, though self-driving cars don't crash into other cars, they "cause" crashes (mostly fender-benders) by driving in a different manner than humans. No evidence that they cause more fender-benders than humans, who drive all sorts of different ways. Apparently, the author thinks this is so obvious, no data is required.
Then, the article seems to propose that self-driving cars should be like humans, who allegedly violate laws in a "safe and principled" way. This ridiculous statement ignores the tens of thousands of people killed in crashes caused by human-driven cars every year, and the many tens of thousands severely injured, by these "safe and principled" violation of driving laws. Most accidents do not involve complying with traffic laws, actually. That's why we have traffic laws -- to prevent this carnage. Individually, however, we still tend to believe that even if we engage in risky behavior like speeding or running lights, we're not the one personally who will crash.
As to some people who think self-driving cars should speed because so many people speed, please check the mortality rates for accidents at various speeds (they're widely published) -- survival rates go down lots as speed goes up, especially if a pedestrian is involved. On the other hand, if you do the math, on most trips the amount of time actually saved is much smaller than the exhilarating sense of going faster than the next guy you get from even moving a couple of mph faster. This is why many major cities are reducing speed limits and taking other steps to reduce traffic speeds. Speeding, the most common rule violation in driving, is neither safe nor reasonable, and the author, supposedly an expert on driving, cannot provide any evidence to indicate that it is. Well, except that he seems to assume that tens of thousands of traffic deaths are acceptable and unavoidable.
Perceptive readers will notice that I haven't cited stats to support my assertions in this comment. That's because it's a comment and I'm not holding myself out as a journalist who's done sufficient research to report on issues of motor vehicle safety. I'm a mere commenter, but be sure, if you do the work to find out what the real numbers are (regarding both driving safety and actual driving times) it bears out what I'm saying.
"Humans violate the rules in a safe and principled way...."
Thanks, it's nice to start my morning with a good laugh.
Maybe the author lives in some magical wonderland filled with good drivers, but this is not what I observe at all. People drift out of their lane for no reason. They careen past the stop line because they can't be bothered to push the brake pedal a bit farther down. They refrain from signaling because they just don't care. They blast through stop signs and red lights without heed to traffic or pedestrians. They brake spontaneously in the middle of a fast moving road. They drive fast in dangerous conditions and slow in safe conditions. They spend half their drive texting because they think they're the one person on the planet who is able to safely text and drive. How is any of that safe or principled?
As a Parisian driver, I think I understand exactly what he means. Other Parisian drivers violate the rules with a nicely predictable scheme where non Parisian may respect the rules but are very dangerous and get horned continuously.
The rules are supposed to be a universal agreement. Otherwise it makes it very dangerous to drive in a different city even though the laws are the same.
To provide some means for censuring those people who flagrantly disregard the unwritten rules (I know that, formally, that looks nonsensical, but in practice it is not.)
The rules are not "for" something in France. They just appear out of committees and political compromises and are often half-assed implementations of nonsensical middle-ground between two coherent positions. Some may occasionally align with public interest.
One of the things you learn living in France is a sense of the rules that make no sense and that everybody is probably disregarding, and the ones that are actually useful and probably enforced.
Always smile to a clerk. They will tell you which mandatory paper is not mandatory and which rule you can break. They will also tell you which rule the state is incapable of following and how to get around it.
It is rather self-serving to say that people following the rules are creating danger. If you are driving in such a way that a dangerous (as opposed to merely inconvenient) situation results from someone following the rules, you are not driving with sufficient caution.
For example, to change lanes, a Parisian will put on his flashing light and change lanes just after even if a car is arriving. He knows that in Paris everyone is attentive to other cars and have seen his flasher. A non-Parisian guy's gonna put on his flasher and wait for space. This never happens and he ends up changing lanes very unexpectedly.
So... when there is somebody who isn't a Parisian and they hit a car because they did not know this, are they in fault?
I live in Paris, and although I do not drive I have the impression that most drivers here are extremely inconsiderate. As an example, there is a crossing which is constantly congested with buses and lorries stuck in the middle here. This is mainly because nobody respects the rules or priorities. When an officer comes in to control the traffic instead of the lights, it gets "magically" cleaned up and the circulation is fluid.
If that driver is signaling, then a move should not be unexpected, especially if the unwritten rule is that one must force one's way into another lane.
You are actually providing an example that the Parisian way of doing things is not as benignly efficient as you suggested in your first post.
If you adopt the Parisian way, you can change lane very easily and quickly even if there is a big traffic jam. I have drive in traffic jam in smaller towns (like Cherbourg), it was more difficult to change lane.
The fact that you can not change lane without forcing is a normal consequence of the increase of the traffic. I think the unwritten rule is that people should not turn on flashing lights too early before acting. The better communication of the intents and a fair level of trust that other drivers will understand it (and not try to block you) makes driving a lot easier (in Paris).
> I think the unwritten rule is that people should not turn on flashing lights too early before acting.
A bunch of highly nuanced rules like this seems very fragile. If you are driving in this manner, you have a responsibility to be alert for things breaking.
This is silly. You're ignoring a reality where a culture has _figured out on their own_ an optimal way to navigate from point A to B in bad traffic, (mostly) safely while constraining themselves to the _intent_ of the law, if not the letter.
There's no better realization of distributed peace. Who is the self-serving person who thinks they are entitled to travel to Paris and drive around without understanding the cultural norms of their unspoken driving rules?
> You're ignoring a reality where a culture has _figured out on their own_ an optimal way to navigate from point A to B in bad traffic, (mostly) safely while constraining themselves to the _intent_ of the law, if not the letter.
Is it a reality? Sounds like conjecture to me, and the fact that many humans are killed in traffic fatalities certainly suggests otherwise.
Driving at roughly the same speed as the rest of traffic is definitely much safer than driving the posted speed limit.
Any other such commonly observed violation would probably have to be studied to answer the question - and given humans' tendencies to take on a fairly uniform level of risk no matter what the rules are, I suspect the question may need to be "does this keep the risk level constants while increasing throughput" rather than "does this reduce risk".
So ... it's plausible both ways so far as I can tell.
Maybe it's rare where you are, but it's common for traffic to flow at greater than the speed limits in many places. Here in the DC area, for example, if freeway traffic isn't so dense that it's jammed up, it's probably flowing at 5-10 MPH over the limit. I usually drive at 60MPH in 55MPH zones and stick to the right lane because I'll be one of the slower cars on the road. Sometimes the difference is so great that 60MPH is still too slow to be safe and I bump it up a bit more.
What one ignores is the traffic that you pass when you're going 65 in a 55. I would estimate probably 1/4 of traffic is moving at or just below the limit. I'm in Houston and we deal with traffic quite a bit and I'm not addressing rush hour bumper to bumper here.
The fundamental question is who is at fault for the differential in speed (i.e. that which increases risk of injury/death with an accident)?
The answer is the person who breaks the law and the one who does so as a precaution to themselves (i.e. following the flow of traffic). An accident by this group disproportionately affects those obeying the law.
So I see three groups.
1. The speeder.
2. The law abiding citizen.
3. The person who is aware of their surroundings and adapts.
Ethically between 1 and 2, 2 has the upper hand. 3, when driving with 1 increases risk to 2. But if 3 joins 2 they increase risk to themselves and to the larger group 1.
France, specifically, has one of the highest rates of traffic fatalities in Europe. Roughly 3500 people die every year on their roads. If those cultural norms resulted in safer traffic, I'd be all for giving them precedence - they don't, though. So now you have to weigh the value of cultural norms around driving style vs. people's lives.
First 3500 is the total amount. You need the number of deaths that were caused by local trafic rules and could have been prevented by strict adherence to traffic laws. You'll be lucky if 5% of those meet that criteria.
You also have to weigh the lives against the value of the time saved -- the whole reason those norms exist. There's no way it comes out in favor of the deaths.
Let's try and ballpark whether it's worth it to speed 5km.
Roughly 410 billion miles driven yearly by the French.
The average speed of a journey in France is ballpark 60km/hr
Average salary is 36k in France with a 1450 work hours a year for €25/hr.
Putting it all together we get €0.42/km or €170B.
Now suppose that everyone always drives 5km over the limit bringing that average to roughly 65km/hr or €.38/km or €156B for a ballpark €14B in savings.
Say you can save 10% of those deaths by driving the speed limit which is extremely generous. I would take odds that not even 1% of the deaths could be traced to speeding 5km over but let's pretend. Even then each life saved still costs €400M.
> You're ignoring a reality where a culture has _figured out on their own_ an optimal way to navigate from point A to B in bad traffic, (mostly) safely while constraining themselves to the _intent_ of the law, if not the letter.
But the intent is not always sufficiently appreciated.
Case in point, urban speed limits at 50Km/h in EU or 35MPH in US. For most drivers (me included), they feel terribly slow and overly precautious. Yet the intent has to do with stopping distance. In those areas, you should be prepared to come to a full stop on a very short notice because a child/dog/ball could be darting out in front of you with zero notice. Not only that but the cars behind you should also be able to stop. Whenever you're driving at 45 in a 35 zone, keep that in mind: not only you need to be ready to slam on your breaks but you must have enough distance with the vehicle in front of you, should they need to slam on their brakes.
Most people only consider the "happy path" scenario, not the "error handling" one.
I suppose, but lawmakers optimize for the 1% error case, actual drivers optimize for the 99% case.
Let me take a thougher position. The value society gains by upping the speed limit on a major road will quickly outpace the value of the lives lost due to accidents. It's just that the injuries and deaths are more visible.
The evidence should be that almost everyone drives at the faster speed anyway and 35mph roads aren't death traps.
> You're ignoring a reality where a culture has _figured out on their own_ an optimal way to navigate from point A to B in bad traffic, (mostly) safely while constraining themselves to the _intent_ of the law, if not the letter.
This is self evidently not true though. Humans are terrible drivers, and the road would be safer if all of those people who are unsafely not following the rules instead followed them.
The reality of the world that we live in is one of terrible drivers that get into lots of accidents.
> culture has _figured out on their own_ an optimal way to navigate from point A to B in bad traffic, (mostly) safely while constraining themselves to the _intent_ of the law, if not the letter.
Exactly. This is why the speed limit on highways is always min("one or two under the misdemeanor speed in this state" "as fast as conditions allow").
Cultural norms have nothing to do with optimal driving. It used to be a cultural norm in parts of the US to drive around with a 6-pack of beer in the console, and one in your lap. Just because a culture has "unspoken driving rules" doesn't mean they're safe or sound.
> Who is the self-serving person who thinks they are entitled to travel to Paris and drive around without understanding the cultural norms of their unspoken driving rules?
Anyone licensed to drive in France, actually. The requirement is that they understand the traffic laws and regulations.
> It is rather self-serving to say that people following the rules are creating danger
Not so simple.
The rules are created via legislation affected by lobbying.
The AAA for decades have been lobbying to keep speed limits low.
In the perfect world, speed limits would be informed by the 85% percentile speed of engineering surveys.
In reality, it's much different.
If a large number of drivers are driving at a speed higher than a prima facie speed limit, the speed limit could actually be low, in which case, the drivers actually following the prima facie speed limit are creating a hazard to the flow of traffic.
OTOH, if the prima facie limit were followed consistently (or enforced effectively), people would not be exceeding it, and speeds would be more tightly clustered and safer.
I saw the same thing in Boston. Instead of "violating the rules", I would say that local drivers obey a set of unwritten local rules that are more appropriate for the situations they encounter. These local rules make things safer and more efficient because the global rules that are written for wide, straight highways don't map well onto low visibility corners, confusing or incorrect signage, low speed side roads merging onto fast thoroughfares, roads (sometimes purposely) without painted lines, etc. I felt that Boston drivers often adopted Right of Way, a notion familiar to boaters. Faster cars with more maneuverability are in general expected to avoid slower cars or those with less maneuverability. If someone pulls out in front of you (perhaps because they were stopped and couldn't find a better time to merge), it's your duty to not hit them. On the other hand, if you are on a side street waiting to merge, it's expected that you wait for a break that's "big enough", then pull out and accelerate as fast as possible. None of these rules are written, but they are still consistent and safer than trying to rigidly apply global rules to the situation.
The roundabout on the Cambridge side of the BU bridge, which I unfortunately have to take home most evenings, is something I can't imagine a driverless car handling. Sometimes it's a single-lane roundabout, but at peak hours it can support as many as five cars wide, eventually merging into a single lane over the bridge, and the traffic pattern is different every day.
Driving on highways is easy, and solving that is still valuable, but it's going to be amazing to see these things work in situations like those.
Tesla know this. Don't forget that every car they make records telemetry and tests new algorithms. It would be great to see Elon stomping around stage like Ballmer: "data, data, data, data, data, data, data!"
There are unwritten rules for driving everywhere and one reason everyone thinks drivers from other states are terrible is that they are not adhering to the same unwritten rulebook.
Back when I commuted more regularly, the traffic around the beginning of September was always particularly awful. My assumption was that school was back in session and that, as a result, you had a lot of people with new travel routes who weren't quite up on the patterns at various rotaries and other traffic chokepoints.
One unwritten rule is - evidently - to attempt to limit speed in the passing lane so that those who wish to pass must do so in the right lanes. Also: brake-check anyone who annoys you.
Well first you're conflating a few things that are totally safe and reasonable corner cutting measures versus things that are actually illegal and should be.
Second, we're not going to replace all cars with autonomous cars overnight, so autonomous cars need to be able to deal with unpredictable human drivers, even if only temporarily.
Thirdly, in my experience, it is oftentimes the overly cautious driver that causes more accidents than anyone else. If the self-driving cars are emulating the people who do the exact speed limit on the highway in the left lane, this is not a good thing because the deviation from the established norms of any given highway is the best indicator for where accidents are going to occur.
Surely it would be ridiculous to deliberately program the machine to break the law, though. It sounds like in the best case scenario, you're describing a situation in which the law should be changed to correspond to how people actually drive.
> Surely it would be ridiculous to deliberately program the machine to break the law, though.
In my driver's ed course, it was explicitly covered that police will often not pull you over for doing 5 mph over the limit. If we're training drivers to casually break the law, then it's only logical we'd be doing the same thing to self-driving car AI.
What am I conflating? Everything I described is unsafe. Note that I'm not talking about failing to signal on an empty street or rolling a stop sign when there's good visibility and no traffic. I'm talking about failing to signal when there are lots of other cars around who could really use that information, or rolling a stop sign too fast to tell if you need to stop, or just without bothering to look.
It's really weird that you use driving the speed limit in the left lane as an example of dangerous rule-abiding behavior. That's against the rules in most jurisdictions, and definitely should be in any where it's not.
I agree that it's possible to break the rules safely. Rolling a stop sign in a situation with good visibility or exceeding the speed limit by a bit is perfectly fine. I agree that humans sometimes break the rules safely. But I disagree with the statement that humans violate the rules in a safe and principled way. When humans break the rules, it's almost never based on any principle besides sloth or selfishness, and while humans do break the rules safely sometimes, it's almost always safe by accident.
The larger point that autonomous vehicles need to not be slaves to the rules is a good one. But humans are not the model to emulate!
> What am I conflating? Everything I described is unsafe. Note that I'm not talking about failing to signal on an empty street or rolling a stop sign when there's good visibility and no traffic. I'm talking about failing to signal when there are lots of other cars around who could really use that information, or rolling a stop sign too fast to tell if you need to stop, or just without bothering to look.
You weren't quite that clear about it in the first post, the rolling stop sign with good visibility was my first mental picture while reading, so I apologize for the misunderstanding. That being said, all of these maneuvers you're talking about could easily cause an accident whether the other vehicle was AI driven or not, so I don't see how it's relevant to the topic at hand: that corner cuts and slights which are technically violations but not inherently unsafe need to be predicted and handled by the self-driving car.
> It's really weird that you use driving the speed limit in the left lane as an example of dangerous rule-abiding behavior. That's against the rules in most jurisdictions, and definitely should be in any where it's not.
Again this was my mental picture, the idea of a robot driving according to the letter of the law reminds me very much of the sort of people who cruise at speed limit in the passing lane, just for different reasons.
> I agree that it's possible to break the rules safely. Rolling a stop sign in a situation with good visibility or exceeding the speed limit by a bit is perfectly fine. I agree that humans sometimes break the rules safely.
Agree.
> But I disagree with the statement that humans violate the rules in a safe and principled way. When humans break the rules, it's almost never based on any principle besides sloth or selfishness, and while humans do break the rules safely sometimes, it's almost always safe by accident.
I would assert that the times they break them safely is not always safe by accident. Every given decision in driving is a balance of risk and reward, the reward more often than not arriving at one's destination faster, and that's where the majority good decisions come from. The bad ones are often not decisions of R&R but of distractions, as you point out like texting, or worse, inebriation. I'm not arguing human drivers are better: I'm merely trying to say that the article is, I feel, attempting to demonize people for driving like people drive, and saying self-driving tech needs to be able to handle the unpredictable nature of human drivers, just as I did on my commute earlier today, if it's going to be a mainstream technology.
I don't see how "They blast through stop signs and red lights without heed to traffic or pedestrians" is in any way unclear.
Of course autonomous cars need to predict and handle sloppy human driving. But that's not what the article is talking about. It's talking about programming the cars to break the rules "in a safe and principled way" the way humans do because humans will expect it. I don't even have a problem with programming the cars to break the rules in a safe and principled way. Where I have a problem is when the article points to humans as a model because that's what they do. It's not.
> Every given decision in driving is a balance of risk and reward....
This is probably the center of our disagreement. Every given decision should be a balance of risk and reward, yes. In reality, for 99% of drivers, every given decision is a balance of reflex, selfishness, and vague remembrance of rules they learned 30 years ago and forgot 29 years ago.
It's true that safely breaking the rules is not always an accident. It's merely an accident 99% of the time.
There's a busy intersection near my house which I cross on foot fairly often. Cars routinely turn right on red without stopping first. Every time I cross that street, I have to carefully watch the oncoming cars until it looks like they're going to stop, because if I just start walking when the walk sign comes on, I will get hit. These drivers aren't risking my life because they're balancing risk and reward. They're doing it because they want to beat traffic from the other road and are too clueless to understand that what they're doing is dangerous.
Again, I have no issue with programming autonomous cars to accommodate flawed human drivers, or even with programming them to deliberately break the rules in cases where it's a net gain. I merely object to the proposal that humans are the model to follow in that respect.
Also understand what author means, coming from Pittsburgh where the roads are simply not meant for cars. There are many intersections so confusing where it takes too much time to even interpret the rules. You regularly find yourself in situations where eye contact and gestures inform your driving more than the rules.
To add to this, there is literally a term "the Pittsburgh Left"[1]. Informal etiquette does really help prevent things like a lane getting blocked for three lights when someone needs to make a left at a busy intersection. However, humans definitely aren't perfect and they likely add a lot of congestion on their own from stuff like following to closely, not maintaining speed, etc.
I feel like this is going to be a huge problem if we have mixed human operated and self-driving vehicles on the road at the same time. If we ever move to entirely self driving cars we could move to some sort of protocol for cars to communicate about ambiguous self driving actions.
It's really common to see the Uber Volvo's driving around, but it definitely causes me to tense up when driving around them. I have no idea why, I realize there is significant risk when around any other car on the road. It's new. Hopefully these issues get worked out in a reasonable time frame, but it is software.
Let's not overlook that once a single self driving car understands these local, informal rules, ALL self driving cars will. That's about a million times better than the current situation where every new human driver has to learn them. I was going to use a number smaller than a million, but 10k or 100k doesn't seem to properly describe the situation. Literally, these systems will be hazards until the day they aren't, and then they will, for all cars, forevermore, be better than every local.
Nor are they easy to redesign. Pittsburgh, Boston, Paris, etc. are generally designed in ways that everyone knows are suboptimal, but the alternative is a catastrophic shutdown in the heart of the city.
It is sometimes amazing how stupid traffic light systems are. One is giving left turning traffic and straight traffic going in the opposite direction green at the same time. That can be solved easily by making sure that all traffic that gets a green light can actually proceed across the intersection. This may need to vary with traffic density, but that's just a technical problem.
The bigger one is to make sure that the next traffic signal down the road is also green. To avoid traffic piling up on an intersection because it has nowhere to go. Of course, this should also depends on the amount of traffic.
Any city can in theory build systems where traffic that wants to go to empty roads gets more green and any traffic that wants to go to full roads gets less green.
You don't think traffic engineers have thought of that? Like, "it's easy to fix this complicated systems problem with no downsides, these idiots just haven't listened to me yet."
I'm fairly sure Boston has intentionally staggered lights to ensure that drivers don't get successive greens in some areas. I would also bet good money that the alternative is some kind of horrific gridlock from emptying side streets too slowly.
Honestly, I'm amazed urban traffic is as good as it is, since the core problems are usually irreducible. (i.e. more people want to go to a place than fit there.) A quick look at gridlock in cities where traffic signals aren't carefully planned and widely respected shows just how bad things can get.
(Suburban traffic is another story, but that mostly seems to be about overbuilt roads with no actual planning at all. My hometown has a couple of intersections that literally don't function - you can get caught at a permanent red light.)
They are a combination of timers and sensors. And yes, we know all about this. Your assumption is correct. Doing this allows traffic to move with greater throughput overall. If it wasn't done, heavier trafficked thoroughfares would have long periods where they were the only priority.
Err... I modeled traffic.
The possible solution would be roundabouts and yield signs, but that would be expensive, require redesign, and we humans pretty much suck at roundabouts.
Intersections are generally a problem because they are constrained by a host of other factors. Complex problem + limited budget generally result in poor implementations. However, in a large enough context you can remove buildings and move roads etc.
So, these things are very much fixable given sufficient priority and resources.
> However, in a large enough context you can remove buildings and move roads etc.
Pittsburgh isn't flat, many times it is not "move this road." It's bore through this mountain, build this bridge, fix the drainage issues around this creek, etc. and fixing intersections almost always leads to more issues downstream.
No, it's the self driving cars. Self driving cars need to learn the actual rules of the road in addition to the written ones before they're ready for mainstream adoption.
Not at the cost of human lives. Yes, self driving cars may get rear ended more frequently etc, but low speed collisions are not nearly as important as pedestrian fatalities etc.
The problem is that you can't viably sell an autonomous vehicle that large, heavy, fast, and operating so close to other vehicles & unprotected pedestrians when it has been deliberately programmed to violate known laws. The makings for an industry-destroying lawsuit are enormous.
The underlying problem seems to be that the rules of the road are broken. We need to ensure everyone drives to the rules, and if the rules are out of date, they need to be updated. Selectively allowing rules to be broken is an extremely bad precedent.
This. If common behavior is against the rules you either need to change the typical behavior or the rules. You can't just wave your hands and say "but I was following the rules". Sure that might save your wallet if you cause a crash but that doesn't scale to the society level.
That works on an individual level but why not modify the rules to fit the norm so the people who exceed the law when practical don't conflict with the people who follow the law reality be damned?
Your comment also reminds me of a proverb to the tune of "bad drivers aren't a problem until they meet".
Sure, one can be that driver that leaves death and destruction in their wake but still look good on paper if the metric is legal and financial penalties since those only loosely correlate with actual bad behavior.
Look, if self-driving cars can't drive like a human then it's already game over. Of course it's possible to completely rearchitect our transportation systems to make autonomous transportation possible. Hell, we could install a series of movable tracks along the roads that send signals to vehicles that would remove virtually all of the difficulty. If autonomous vehicles aren't feasible without sweeping changes to traffic law then they're not feasible yet.
which begs the question, will self driving cars be coded to break the law, specifically speed limits. How do we accommodate that into regulations governing their use. We have a hard law that states, speed limit. Yet "everyone" knows otherwise.
I am curious how we sort this out. Will roads become stricter with cars self enforcing the limits? How do self driving cars know who is next at a four way stop?
There is not a hard law that states speed limit at least in California; instead the law sets 'reasonable and prudent' as the limit, and that lawfully set and posted or legislated limits set forth evidence of what is reasonable and prudent. Driving over 100 mph is separately off limits, however. (I would rarely find that speed to be reasonable or prudent!).
A self-driving vehicle would likely be recording evidence of conditions which could show that the speed was reasonable and prudent given the conditions, which include other traffic on the road.
The basic California speed law is:
> 22350. No person shall drive a vehicle upon a highway at a speed greater than is reasonable or prudent having due regard for weather, visibility, the traffic on, and the surface and width of, the highway, and in no event at a speed which endangers the safety of persons or property.
It is furthered by:
> 22351. (b) The speed of any vehicle upon a highway in excess of the prima facie speed limits in Section 22352 or established as authorized in this code is prima facie unlawful unless the defendant establishes by competent evidence that the speed in excess of said limits did not constitute a violation of the basic speed law at the time, place and under the conditions then existing.
Additionally section 22348 describes a specific additional infraction for driving at a speed greater than 100 miles per hour.
I really wouldn't want to program a car that obeys speed limits in 49 states, and in one state "record[s] evidence of conditions which could show that the speed was reasonable and prudent given the conditions"
From an implementation perspective that sounds like a lot of overhead just to let Californians drive fast.
Autonomous cars will need to be able to determine a safe and prudent speed in every state. I'm pretty sure every state has laws saying that speeds must be lower that the limit in bad conditions, and even if they don't, the car will have to do this just for basic safety. It shouldn't need any special casing for California.
About a quarter of US jurisdictions follow the same basic rule as California, where exceeding a posted speed limit is prima facie, but rebuttable, evidence of a violation.
So it's not a 49/1 rule.
And, in any case, 100% of US jurisdictions prohibit driving at a speed above the safe and reasonable speed if that is lower than the posted and/or statutory limit, so you need to be able to determine what is safe for conditions anyway; it's just a minority of states where that is a relevant determination when the results are above the posted limit.
You have it backwards. Everyone is going to drive fast regardless of the law. Cali is just a state that recognizes this. Writing the law for humans should be celebrated. You can either take that into account or you close your eyes and plug your ears to reality and release a shitty autonomous car.
You may be underestimating the market of Californians who want to drive fast. For a long time, to sell a car in California at all, you had to put a special emissions package together, with was a lot of overhead, but manufacturers still did it, because losing the California market to save overhead doesn't make sense when selling cars.
The rest of your statements are pretty well founded but
> There is not a hard law that states speed limit at least in California
Incorrect.
I don't know what you mean by "a hard law", but, in California, there is a section, from the very statues you present, that sets a statutory maximum limit.
The complication, being that there are three maximum speed limits in California: 55mph, 65mph and 70mph, depending on the scenario.
CVC §22349 sets a maximum speed limit of 65 mph throughout the state (except for the 70 mph limit on some freeways).
CVC§22349(a) reads: Except as provided in Section 22356 [70 mph on some freeways], no person shall drive a vehicle upon a highway at a speed greater than 65 miles per hour.
The CVC§ 22356 referred to above allows a 70 mph speed limit if justified by an engineering and traffic survey where a 65 mph limit would otherwise be applicable.
Try arguing the 22350 while you have been cited the §22349 on a ticket. Not going to fly regardless of what one might read on websites online.
Would you accept "humans violate the rules in a safe and principled way, plus lots of other ways"?
I mean, you're obviously correct, human drivers do idiotic and deadly things all the time. But I think we can distinguish that from the behavior in the article.
Gently rolling a stop sign at a high-visibility intersection sounds like a safe and principled violation. Driving around a car waiting to turn left is a bit less safe, but still a principled or predictable violation. The Pittsburgh left is a (fairly) safe and principled violation.
None of those things are comparable to texting and driving, or ignoring a red light. They're informal adjustments to the traffic code that other drivers really do anticipate, and which aren't nearly as dangerous as general "bad driving".
I'm not saying it is. A rolling stop at a busy intersection where you barely miss other cars and/or pedestrians is insane behavior, though, and that's what I'm talking about.
Yes, there are people who carefully evaluate the situation and only roll their stops when it's safe. They make up a minuscule percentage of the population, though. Most people just do it out of laziness or selfishness, and when they do it safely, it's a happy accident, not a deliberate policy.
As my driving instructor used to say, "Stop signs and red lights do not actually reach out and stop the other driver."
The lesson: just because it's your turn to go does not mean that others will not enter the intersection when it's not their turn. I agree that rolling stops at a 4-way stop are generally safe when you can clearly see that no other cars are at or approaching the intersection. But in my experience, the fact that the other person is at fault has been little consolation for having to deal with a totaled car and a permanent backache.
OK, obviously that's true. But if it's empty, you have good visibility, and you plainly have the right of way, well, I'm just going to take it on faith that nobody is hurtling down the road at 80 mph. You can't possibly account for everything; that's the risk of driving.
Right, so the reward/risk isn't enough for you. It's also prudent to stop at a green and look both ways but nobody does this either.
At the end of the day you either trust other drivers or you don't. You're not exposing yourself to any significant addional risk by making these 'safe violations' than you do driving in general.
But in some sense you do, because you don't come to a dead stop at every intersection, even when it's green. And it's a bitter pill to swallow, but no amount of defensive driving can make you totally safe. Getting rear-ended sucks.
Sure, but you can take some relatively easy actions to significantly reduce your chance of injury by a crazy driver. I don't come to a stop at green lights, but I do glance both ways to see if there's oncoming traffic. Occasionally there is: I've seen people blatantly run red lights.
OK, sure, but unless you have some extremely fast pedestrians where you live I don't see the relevance to the discussion of whether it's reasonable to do a rolling stop.
Like I feel like I'm talking to people on other planets right now. It's like you've never seen anyone do a rolling stop before. This isn't someone careening through an intersection at full speed. This is someone too lazy to actually bring their car to a complete, thud, stop and gently roll into the intersection at like 5 miles an hour. You're not risking pedestrian lives, you're risking a slightly more abrupt brake if there happens to be unexpected pedestrian or cross traffic.
I never do rolling stops. I've been in situations at night where, if I had done a rolling stop, I would have gently hit someone on a skateboard or bicycle. Some people completely ignore vehicles at crossings. It almost feels like they're trying to get hit by a car.
Actually 5mph is quite fast. When I do a rolling stop in my car, it's typically on the order of perhaps 2mph, which is a speed where I can proceed in 1st gear without coming to a complete stop and then having to start again from a stop.
I just ran the math, the kinetic energy of a 2000kg car at 5mph is the equivalent of the kinetic energy of a 100kg bicycle+cyclist at 22mph.
I absolutely roll stops, but it's important to recognize what we might miss. My old commute had a series of 5 (virtually useless) stop signs in a row, an arterial with very little cross traffic but tons of 4-way stops. (should have been roundabouts). The routine of the drive, and the necessity to stop, the focus on the stop signs and stop lines, and the almost total lack of pedestrians meant that I'd occasionally start off before noticing the cross traffic or pedestrian. (yes, this is more of a boy-who-cried wolf kind of thing). Also, modern cars are amazingly difficult to see out of-- pedestrians can be hidden behind thick airbag-containing A pillars. When you're moving, however slowly, everything in your field of view is also moving. When you stop, it's much easier to spot objects that are NOT stopped.
Hell, just last week I was making a left turn on my bicycle. As I approached the intersection, I slowed, looked each way up the road several times as my sightlines improved, and just at the moment I committed to my turn and began to proceed, I saw the minivan coming from my left that was on a collision course with me; I locked up my brakes and stopped just in time. I know I looked exactly where the van was coming from, multiple times, but somehow missed it. I suspect, again, our brains are much better at seeing what's in motion when we are ourselves stopped.
Drivers have vastly different reflexes and visual acuity.
I have been running for 25+ years and have on occasion been traveling at close to 10 mph and usually at least 8 mph. So California rolls are pretty much the bane of my existence.
This is primarily because the "roller" may be planning to take a right turn and are looking straight and glancing left and don't even look my way traveling from their right.
I do take responsibility for my own survival so I wear lights, and try my best to make eye contact before entering intersections etc. but I have had many close calls where they did not even see me due to their own target fixation.
Now that I am on a bike more often I feel even more insecure because my speed is so much higher.
I'm sure that every rolling stop you do you're utterly alert to potential obstacles from every direction, but I see plenty of "rolling stops" come far too close to doing damage for comfort, and it's usually damage to a road user who has the right of way.
It's easy to bitch about people not following the letter of the law when you're not spending 45min in a rush hour commute every day and the benefit of not following the letter of the law is not tangible to you.
On the contrary, I'm a lot more uptight about it when I drive frequently, because I eventually get tired of people nearly killing me or breaking my stuff because they can't be bothered to put the least bit of effort into their driving.
You do not even notice the small ones, which happen all the time. How many people cross solid lines? How many people sometimes drift a bit out of their lane? How many people scrupulously follow the speed limit like a computer does, not even accidentally drifting above it? How many people simply screw up the order of operations at a busy stop sign? How many people always scrupulously use their turn signals in accordance with the local laws? (You may use them frequently, even scrupulously, but I bet you often turn them on what is legally too late.) I've been driving through a construction zone all summer. People definitely don't drive "45 when workers are around". They don't slow down when the workers are behind the concrete barriers and a dozen feet from the road, and may drive 25 when they're sitting right next to the road with just a barrel a bit between you and them.
Even a cop tailing you and desperately hunting for reasons to pull you over still probably lets through plenty of violations. There's too many laws for the cop to keep in their head.
Crossing solid lines is not illegal in my state. Drifting a bit out of your lane is pretty dangerous. Usually you get away with it, but if two people do it at the same time while adjacent, you're going to bump. Do it when there's oncoming traffic in the next lane (which I see all the damned time) then you could easily get people killed. Mildly exceeding the speed limit is fine, but people don't do it in a carefully considered way because they know it's still safe, they do it because they're in a hurry and know they probably won't get pulled over.
Crossing solid lines (or crossing a gore, at least) is illegal in my state.
Meanwhile, a lot of highway on-ramps are painted with single-lane-width, high-angle merges that are dangerous to adhere to - they leave you making a ~30 degree turn while accelerating into traffic. As a result, everyone starts their merge early in the gore area, matches speed, and merges at a gentle angle.
(The road designers presumably know this, and keep the marked path narrow regardless so that drivers won't attempt to form multiple lanes.)
This is a pretty great example of what I'm describing - I don't think people make great decisions about violating driving rules, but I also don't think it's fair to call safe decisions "inadvertent". Not breaking that law is genuinely dangerous, and people violate it specifically to be safer.
I think you're attributing way too much thought and care to the average driver. I see people merge early like you describe all over the place where it's completely unnecessary. It's probably not a conscious choice for safety over obeying the rules, it's just random idiocy that happens to work better in this case.
I think it's only contextually illegal - on narrow streets, you might have to enter a bike lane, drive in a parking zone, or otherwise break a rule to have room to pass.
It's called a Pittsburgh Left?! When I first came to the US to Boston it drove me crazy, to me it was the epitome of uncivilized American driving "how dare you cut in front of me without my permission!"
But it makes sense at the right kind of intersection, and of course I adopted it because otherwise I'd be the odd one holding up the traffic making people frustrated.
>how dare you cut in front of me without my permission
Allow me to fill in between the lines:
"What is this, some sort of (mostly still) free country where people are not given harsh punishments for all behavior that prioritizes the individual over the group?"
"By accepting a modest delay in going straight, the opposing traffic has saved the left-turning traffic waiting an entire light cycle to turn left, as well as saved an equivalent amount of time for all the cars that otherwise would have been stuck behind the left-turning car. In situations where there is so much oncoming, straight traffic that a left turn would not be otherwise possible during any part of the light cycle, the Pittsburgh left can allow a line of left turning traffic to proceed incrementally."
It's not the safest bit of driving, but it's enormously pro-social. As that quote suggests, the problem isn't "I want to turn left" but rather "there are 30 cars behind me going straight, with no room to pass me on the right".
If people just wanted to turn left, this would happen everywhere. Instead, I see it almost exclusively in cities with narrow streets lacking left turn lanes. In extreme cases, it's even personally beneficial for the yielding driver - the alternative is sometimes gridlock that circles back around to them.
None of which makes it safe, but it's pretty obviously common for reasons other than selfishness.
I had no idea that 1. this had a name and 2. it was illegal. I've done that Left so many times back when I lived in Pennsylvania, I figured it was totally acceptable. I've been in situations stuck behind a left-turner for 5+ light cycles because he couldn't/wouldn't turn left due to unrelenting oncoming traffic. What's the legal solution? You can't wait forever!
(We don't seem to have this problem out here in California, where every street has at least four lanes!)
As far as I know, the only legal solution is that one car gets to enter the intersection and then turn left after the signal has gone red. Of course, that's often a terrible answer - clearing one car per light cycle is a route to spectacular gridlock. (And with Boston drivers consistently running newly-red lights or blocking intersections, you sometimes don't even clear one car per cycle.)
Wikipedia does hint at a legal kludge, though - "I judged oncoming traffic to be moving so slowly that I could safely turn". The idea I guess being that it's not a change to driving rules, it's just slow acceleration and a legal turn.
I can't imagine that would hold up real well legally, especially if people are flashing headlights and making an invitation. On the other hand, I've never heard of this leading to a ticket in Boston - it just can't compete with the other violations that happen constantly!
> Gently rolling a stop sign at a high-visibility intersection sounds like a safe and principled violation.
If you do this yourself, please reconsider. My experience is that drivers who do this (especially at 2-way stops) are watching carefully for perpendicular traffic while blithely ignoring pedestrians approaching the crosswalk that they are rolling through. As a pedestrian, I consider this the second highest threat to my personal life+limb, after left turners who (again) watch for cars and not pedestrians.
Yes! Drivers often seem to only care about the possibility of colliding with another vehicle. So by default they pull right through the crosswalk and/or stop line in order to maximize their visibility of cross traffic and maximize their speed of making the turn without hitting another vehicle. Their actions are both selfish and passive aggressive toward pedestrians, yet at the same time their actions are also completely _rational_ when the driver is ignorant of (or lacks empathy for) pedestrians. They get away with it because it's such a small violation of the law and because pedestrians take on the responsibility of not getting hit.
Yup. If you want to make roads safer for non-drivers, make design choices that make drivers feel more threatened. Turn one-way avenues into two-way streets with street parking, etc.
I see so many entitled, selfish pedestrians downtown, every day it's not even funny. People keep acting like drivers are the only people causing problems, but if pedestrians would quit crossing after the DO NOT WALK, a ton of collisions would be avoided there too. Pedestrians forget to check for, or blatantly ignore cars all the time. I can't count the amount of times I've avoided hitting some pedestrian, because I had to take the responsibility to look out for them.
If cars teach us anything it is that having a massive sense of entitlement is an effective strategy for promoting your own interests. That is why cars are treated with priority in so many places and can delegate responsibility for safety onto other users.
If you introduce the danger then you should take responsibility for safety.
The law puts most of the burden on drivers quite intentionally, and I think properly. My general impression is that the average driver doesn't understand the law and don't appreciate the extent to which it is designed to favor pedestrians.
To give one example, drivers are required to yield to pedestrians in cross walks and to pedestrians approaching crosswalks while pedestrians are only required to "exercise due care," which is a much weaker requirement. If you want to see how weak this requirement really is, look at the case law. In many cases in which the average driver thinks the pedestrian is in the wrong, the law disagrees and would place the lion's share of blame on the driver.
> To give one example, drivers are required to yield to pedestrians in cross walks and to pedestrians approaching crosswalks
Is that actually the case? I know that the former is true, but I haven't come across a law stating that drivers are required to yield to pedestrians who are not already in a crosswalk. Does case law differ from the text of the law it refers to?
> while pedestrians are only required to "exercise due care," which is a much weaker requirement. If you want to see how weak this requirement really is, look at the case law. In many cases in which the average driver thinks the pedestrian is in the wrong, the law disagrees and would place the lion's share of blame on the driver.
I would be curious about what case law would say about a collision like this: https://www.youtube.com/watch?v=io_XBK0twtQ. Specifically, who would be found at fault/liable.
Nobody's acting like drivers are the only people causing problems. However, drivers seem unique in their ability to feel that they're doing everyone else a favor by allowing them to cross THEIR roadways.
Heavier, faster-moving vehicles are typically expected to yield to slower ones, whether on the roads or on the local multi-use path, because they're the ones likely to cause the most damage when a collision occurs.
I did this when I lived in a rural setting and there were no sidewalks or crosswalks. In that case, I'm talking about rolling stop signs at entirely-empty intersections, where you can see that there are no cars or people around. At that point, I'm not sure I feel so bad about it - seeing a person at all is an adequate cue to drive differently.
I don't anymore, though, because I live in a city. Noticing everyone and working out their intentions means actually stopping.
Which, I suppose, is the sort of difference we might get between a self-driving car and a responsible human driver; the car isn't likely to treat stop signs contextually.
The presence of a stop sign is usually the result of a determination based on either engineering principles or (often fatal) experience that the intersection does not provide adequate conditions to be safely rolled, whereas if no such determination were made, either an uncontrolled intersection or Yield would be present instead.
It's true that a self-driving car will likely not be designed to substitute it's judgement for those of the authorities placing signage, though I'm not convinced that this is a bad thing for safety, or that the predilection of some human drivers to disregard such signage is a good, or even harmless, thing.
> The presence of a stop sign is usually the result of a determination based on either engineering principles or (often fatal) experience that the intersection does not provide adequate conditions to be safely rolled
A lot of times, they will be placed as a form of speed control. The MUTCD has requirements regarding the use of stop signs, but they don't appear to be followed most of the time. Otherwise, we wouldn't have nearly as many stop signs at intersections where there is plenty of visibility of cross traffic.
Of the two parts pertaining to complying with a stop sign (first stop, then yield), the second is far more important than the first when it comes to avoiding a collision. You can certainly come to a full stop and subsequently fail to yield and then collide with a pedestrian or vehicle. But even if you don't come to a full stop and do yield to a pedestrian or vehicle that has the right-of-way, you're not going to collide with them.
> A lot of times, they will be placed as a form of speed control.
An interesting thought: what's the lowest speed limit sign you've ever seen?
I've seen 5 MPH in parking lots, but never on an actual road. For official road signs, I virtually never see anything below 15 MPH. I'd be pretty surprised if this was actually the ideal driving outcome. There has to be some intersection at which 5 MPH and a yield sign is the ideal arrangement.
But given how traffic laws are enforced, and how people tend to ignore low speed limits, it's not a very practical arrangement. Instead, stop signs go in everywhere that very low speed limits are required.
(I'm still vaguely wondering whether a lot of people in this thread have ever driven somewhere very rural. Gently rolling a stop sign at an intersection with a mile of visibility and no one else around is not some wild risk.)
The Pittsburgh/Boston left is extremely dangerous if you're not prepared for it to happen. I can't find a citation at the moment, but I believe in Ontario it constitutes stunt driving, and carries fines comparable to street racing.
Although I'm in favour of the law in general, I really dislike the 'stunt driving' name. I'm not sure that many (or any) of the offences covered under the law really qualify as stunts. Dumb and dangerous, yes, but not stunts.
Call this law something else, and save 'stunt drving' for people who drive their cars on two wheels and other related craziness. :)
Despite living in the Boston area I don't really do this, even though I see others doing it, because, well, I don't really trust the other guy to stop.
Are you sure you live in Boston? Usually, the goal seems to be to dare the other driver, sometimes coupled with taunting.
Yes, people do drive differently in different regions. Boston is the only place where I have had two cars totaled. It was parked both times. The second time the driver of the other car seemingly backed up and smashed into it again.
I have done some traffic modeling in your area, and lived in the area. Boston has some aggressive drivers but that's probably due to the infrastructure being based on things like cow paths.
Eh? I'm not in CA, and like the other answers I see this everywhere in the country. I suppose it's rarer in places where intersections are almost never quiet and empty, but I've seen it all over New York (not NYC), Ohio, Wisconsin, and a half-dozen other states. It's standard practice anywhere rural, and in a lot of suburbs.
I wouldn't be quick to assign blame. Most of the lateral parking spots can only be occupied by driving backwards, and drivers don't usually leave enough room for the one in front to make the maneuver. If the car in front of you is signaling right (meaning it's looking for a place to park), please stop before the parking spot.
Sure I do. And I often get it wrong because people are extremely unpredictable.
I'm coming up to an intersection and I see someone edging out, so I slow down and prepare to panic stop in case they pull out in front of me. They don't, so I proceed. That's good defensive driving, but a failed prediction.
I see someone with their signal on in the next lane, so I back up and give them some room to move over. Turns out they just forgot their signal was on. Again, good defensive driving, but an incorrect prediction.
People don't break the rules in anything like a consistently predictable manner. Not around here, anyway.
Predicting someone "might" do something is not a prediction. You would have to predict that they will pull out in front of you or that they won't.
Noticing some behavior and thinking of all the possibilities is not predictive behavior. What you are doing is making a reaction based on the action by another driver.
In one case, you notice a driver edging forward, so you slow down in case they pull down. If you were predicting that they were going to pull out, you wouldn't just slow down, you would slam on your brakes.
In another case, you notice a driver with a blinker on, so you assume they are changing lanes, and slow down to give them additional space. This one is closer to a prediction, but in reality a driver would notice the blinker before they get anywhere near the car in question and would have plenty of time to tell whether it's a forgotten blinker or a conscious attempt at changing lanes. Or they don't see it from far away because traffic is moving very fast and/or is very heavy, so they just keep going making the prediction that the car w/ a blinker on will wait until the lane is clear (aka after you pass them) to change lanes.
>Maybe the author lives in some magical wonderland filled with good drivers, but this is not what I observe at all.
The author of the article has not said those words. They were said by Karl Iagnemma who is a scientist from MIT, and the CEO of a company that makes software for self-driving cars.
It didn't sound like something a reasonable person would say so I tried to track down the context of the quote, and it seems to be a mystery. My guess is that the quote came out of a larger interview, and it obviously meant to highlight the fact that 'standing out' by following rules in a way that other drivers are not expecting you to, makes things much more worse. That makes sense to me, as opposed to your interpretation.
In other words, bad drivers tend to run into things. It isn't really the case that no one ever conservatively follows the rules of the road while driving manually.
So, in other words, the law is wrong. And the is kept as such for purposes of revenue enhancement (illegal taxation)?
I can see a few interesting govt things coming out of "Autos":
1. Revenue from bad laws drops to 0. These vehicles follow the letter of the law. And they have the logs to prove any sort of claim against the former.
2. A false claim is filed (BLM makes about a point about this..) and says driver did X. Is this the person's responsibility, or the company who wrote the AI?
> the reality is that the self-driving vehicles are overly cautious.
Humans contribute to lots of accidents when they unconventionally yield the right of way. Other drivers which should be yielding are now faced with the unpredictability of the too-cautious ("dangerously" cautious IMO) driver re-claiming their right of way. Or they face the problem of other drivers in other lanes who don't notice that the incorrectly-yielding driver is letting another vehicle into their right-of-way.
Inappropriate yielding should be penalised just like lack of appropriate yielding.
At sea the COLREGS require ships with right of way to maintain their course and speed (unless of course they have to give way to another ship, or make evasive maneuvers). Of course fishermen tend to ignore this rule, changing their course and speed whenever convenient to them.
I'll typically just sit there until the person yielding their right-of-way gives up and goes. No way I'm pulling in front of someone with the right-of-way only to have them change their mind on a whim. I just assume they're trying to commit insurance fraud and ruin my day, whether they are or not.
This is my tactic as well. If I can't trust them to know they have the right of way and use it, then I won't trust them enough to put myself in front of their vehicle. That goes for driving as well as walking.
I would argue if an accident occurs from your example, none of the parties involved were yielding properly to begin with. There shouldn't be an accident (nomenclature is more correct with "collision") with a yield.
The worst-case scenario when yielding properly, with unconventionally given right-of-way is that both parties are at a complete stop and have angry drivers behind.
Now, if you are disobeying the law and are following too closely and collide with someone in front of you, that's a different situation......
Yeah, my grandmother does this kind of thing all the time (or did, back when she still drove). Violating expectations out of an abundance of caution causes a lot of accidents and near misses.
I'm sure that when you're trying to engineer a participant in a complex system, factoring in your own impact on that system and how it might come back to affect you is as simple as it sounds. I mean, it's not like self-reference is the generator of intractable complexity in every situation it is found, right?
I think most people discount the effect that connected cars will have on safety even when you have mixed levels of human operated and level 1-5 AVs. If a car operated by a human has even a basic connected control system in it then a AV can simply communicate that the driver is about to hit the AV and the human operated car will take some basic corrective action.
The interesting tech is not only AV but the infrastructure where cars communicate automatically about which lanes there in, what their intentions are etc.
The current use case I think is cool is a car comes up to a slow truck and the truck communicates to the car that the passing lane is clear.
> The interesting tech is not only AV but the infrastructure where cars communicate automatically about which lanes there in, what their intentions are etc.
I don't think that this will really work. Considering how interaction between tech of different manufacturers works in other areas, I doubt that cars will be able to safely communicate within milliseconds in high risk situations. But they won't need to, they can just use their own sensors to detect what happens around them.
"One consideration in implementing the ship's autonomous functions was making its movements seem human-like. Any maneuvers it makes to avoid collisions have to be detectable by crew on other boats, so changes in direction (which are more easily spotted) are preferable to changes in speed that might be just as effective but are less obvious signals."
Obeying rules is nice but you need to account for behaviour of human actors as well.
If autonomous cars become more common this seems like a problem that would solve itself. If more cars on the road are following the letter of the law then people will stop being so surprised to find a car following the law and won't be caught by surprise.
Exactly. And then there will be a tipping point where autonomous cars are the norm with people 'riding' in them and all of our lives will be made easier because autonomous cars will do the driving, and the traffic laws pretty much will be obsolete. It will just be routing autonomous cars various places.
At first (and probably for quite a while) self-driving cars probably need a big sticker that's easy to see so people know that they are going to be "different" on the road.
People will rapidly learn to recognize those "damn robot drivers" and adapt the same way they do now when they see the "student driver" placard or the unmistakable sway and surge of an intoxicated driver.
I wonder if having the sign the same colour as the L plates but with an A would help, given that learners are often much closer to following the rules and the colour might help human drivers pattern match even if they're not really aware of the A plate yet.
One of the many ways that drivers can be more predictable to other drivers is to disobey the laws regarding pedestrian rights of way and safety. Drivers not only disobey these laws all the time, but they expect you to do so too, lest they honk and use their car as a weapon to show you how angry they are that you let someone cross the street at an unmarked crosswalk.
Unfortunately for all those with fantastical fantasies about our autonomous car future, when they finally get here they’re going to be slow. And that’s a good thing, even if it means the utopia is still slow and congested.
Drivers expect other drivers to follow the rules of the road, not to be nice.
Did you hear about the woman in Canada [1] who tried to "be nice" and stopped in the middle of a highway to let some ducks cross? She killed a motorcyclist and his passenger.
Don't be nice -- be predictable.
Same is the reason why the bots get rear-ended -- everyone expects them to roll through a stop, who in the world does a complete, 3 second stop?
A crosswalk by definition is a place where pedestrians can cross with the right of way. They absolutely can be and often are unmarked, and exist in practically every first world country, including yours.
Anytime a vehicle gets rear-ended, fault belongs with the rear-ending vehicle’s driver for either failing to maintain safe following distance or failing to pay adequate attention to conditions in front.
And rolling a stop is not following the rules of the road, it is a dangerous violation.
> Anytime a vehicle gets rear-ended, fault belongs with the rear-ending vehicle’s driver for either failing to maintain safe following distance or failing to pay adequate attention to conditions in front.
I don't think that's true on a highway or similar road. Stopping in the middle of the highway is very dangerous. You cannot blame the 150km/h+ cars to be able to stop safely in all situations.
> I don't think that's true on a highway or similar road.
It's definitely true on a highway (which is anything that's not an alley), and even a freeway (which is probably what you meant.)
> Stopping in the middle of the highway is very dangerous.
Failure to maintain adequate following distance on any road is dangerous; encountering stopped traffic or needing to stop rapidly, even on freeways, is a regular occurrence.
> You cannot blame the 150km/h+ cars to be able to stop safely in all situations.
I don't blame cars for anything, I do blame drivers for failing to maintain a safe combination of speed and separation.
And there's virtually no road, freeway or otherwise, where 150km/h+ (93mph+) is a safe and reasonable speed for any driver, irrespective of separation, so I can quite easiky blame a driver for choosing to drive at such a speed in the first place.
Absolutely you can always blame the car that does the rear-ending, it is your responsibility to maintain distance and speed such that you can stop if the car in front of you slams on the brakes. Which can happen for a multitude of valid reasons, like something dropping of a truck in front, or an accident.
You will absolutely be at fault and stuck with the bill in every single rear-ending case, good luck arguing otherwise.
She was sentenced in December 2014 to 90 days in jail to be served on weekends, three years' probation and 240 hours of community service, and given a 10-year driving ban.
That's not the Supreme Court of Canada...that's the Quebec Appeals Court.
Regardless, it was dumb of her to stop there. But nowhere near as dumb as your idea that crosswalks are marked by definition. If you're going to give driving advice, don't be wrong. Especially on something like that, where you could easily kill someone and where the penalty could be as severe as manslaughter.
I think a big part of the problem is that there is too many vague, unspoken rules of the road. Making them explicit would also be impossible because there would be far to much to calculate accurately while driving (for the average person).
For example: corner to corner crossing is a normal law for pedestrians. Technically, you should always yield the right of way for pedestrians to cross in this manner. Say you are on a two-lane main road through town, and there is tons of traffic. Do you stop at every single corner to let one person cross?
The law says yes. Everyone waiting on you is going to say HELL NO! The people that stop at every corner think they are god's angel doing the right thing, but really they are being a massive sh*t head.
That's just one example of the fuzzy line between breaking the law for the common good or following the law to a tee to the detriment of everyone on the road.
Obviously, the best solution is just better city planning so that pedestrians rarely have to interact with traffic. When they do have to interact with traffic, it should be limited and explicit ( red lights, big marked crosswalks, etc).
> That's just one example of the fuzzy line between breaking the law for the common good or following the law to a tee to the detriment of everyone on the road.
Except your “common good” isn’t common. It benefits drivers by mere seconds in exchange for a mortal risk to pedestrians.
Like it or not, but the law is written as is for a reason: pedestrians are the lowest common denominator of society. Everybody is a pedestrian some of the time. When your expectations are for pedestrians to go several minutes out of their way to find a light and crosswalk so they can cross the street, just so you don’t have to slow down for a split second, it is you that is being the complete shithead.
Luckily autonomous cars will soon make all of the enpathy-deficient sociopathic shitheads wait their turn, like they should have been doing all along.
It doesn't impact drivers by mere seconds. It creates traffic backups all the way across the city. When a driver waits 3 blocks in a row for pedestrians to cross, it creates a backup of probably 30 cars. This is my daily experience in my city.
I don't think pedestrians should go way out of their way to cross, I think they should just wait for gaps in traffic.
Likewise drivers should use good judgment and stop for an old lady, but keep going if it's an in-shape jogger.
I take this as evidence to support my argument that (at least in North America) stop signs are mostly unnecessary and utilized badly. 90% of them could go, some replaced by yield signs, some just removed. An intersection with a line of sight onto cross traffic should not have a stop sign.
Maybe the robot drivers will finally drive home how dumb some of our road infrastructure is.
This is purely an anecdote, but in my city it's hard to cross a road as a pedestrian even when there's a stop or a priority light for pedestrians. I've been honked and yelled at for "not crossing fast enough" or whatever. Of course it's not the norm and most drivers are respectful and doing just fine, but even a couple of those in a month gets annoying.
I shiver at the thought of having to cross any of these if there was only a yield sign. The principle that I have priority is not important enough over not getting hit by a bad driver.
TL;DR: I agree 100% with your point, but I do think bad drivers should be taken into consideration.
You're definitely correct that yield signs make it harder to be a pedestrian–most American cities are awful for pedestrians to begin with, outside of a small core downtown area. Even intersections with proper crosswalks and lights are often massively wide with no middle refuge, have traffic crossing the crosswalk from multiple directions/lanes and poor sight lines, require multiple long waits at lights to cross all the various lanes etc.
There seems to be a growing use of traffic circles (with yield signs) in the US, which are very much like playing real life Frogger as a pedestrian: cars coming quickly around the circle, often not 100% comfortable with how it works, watching for cars around them instead of paying attention to people crossing at the exits. In a busy intersection you sort of have to step out and try to get someone to stop for you or dash across a gap in traffic–a firm stop required for cars makes for a much more pleasant crossing.
A trivial sign that you find a minor inconvenience probably saved some innocent driver thousands of dollars in insurance payments when a separate, at-fault driver took them to court.
I used to feel the same way until I realized stop signs are traffic controls that do more than cause cars to stop. They reduce traffic speed, allow pedestrians to cross roads, and can shift traffic patterns. Often these secondary effects are not obvious to drivers.
There are stop signs that shouldn't exist but they're the minority.
How much is this exposing humans who are not paying attention (ie: on their phone) while driving?
I would like to know to what extent are accidents caused by robot drivers being "odd" versus humans not paying attention and then blaming the robot for being odd.
Lol. I love how they say humans need to drive more like robots when it comes to traffic cameras, insurance, and making big money for insurance and gov't.
However, when it comes to something that threatens these industries...it's suddenly the humans that need to lighten up?
I'm ready to get rid of the entire auto industry once and for all, along with all of the insane amount of income they generate off the backs of regular people.
In Australia, stop means stop. If it’s safe for you to roll through, there will be a GIVEWAY sign. But if you roll through a STOP you’re an absolute fool who’s risking the lives of yourself and others. We don’t have four-way stop signs. We have roundabouts, or give one street priority and the other STOP signs in cross-streets.
I recently passed my California driving test, and I’ve observed that the placement of STOP signs here does not correlate with safety in the same way as Australia and lots of people roll through them.
I think self-driving cars would do a lot better in Australia.
You've hit on what I see as the big problem with stop signs in the US - they are inconsistent. Sometimes they are used at intersections where failing to stop is very dangerous, but much more often they are used when a yield sign would be sufficient. It conditions people in their normal driving to think of stop signs as recommendations, not requirements. This would seem to increase the danger of intersections where stopping fully is really necessary since people are more inclined to roll pass the stop sign.
There is an interesting tangent to this story that I think would answer some of the written rules of the road versus custom. By and large, do the people designing robot cars like to drive? Do they like getting out on the road and driving around?
372 comments
[ 4.0 ms ] story [ 285 ms ] threadI think it's incredibly unwise to program a machine to "strategically" break the law.
Yeah, that's definitely a worthy goal to aim for.
Genuine question: what do you think is the better alternative to the 100% complete stop? I'm not saying I think there is or isn't a better alternative, just wanted to know what you had in mind.
I think that would be to slow enough that you can stop if you need to, but not actually stop completely --- something which almost everyone already seems to do, because it saves wear and helps with passenger comfort too. (I've been in vehicles where the driver insisted on completely stopping exactly as legally specified, with the accompanying double-jerk of deceleration followed by acceleration. Not pleasant at all.)
You can do that even if you're braking very hard, and it doesn't have to make your stop slow. All you want to avoid is that bounce back right at the end, and backing off the pressure a bit just before the full stop will be enough to achieve that and create a very smooth ride. Of course you will always have the deceleration but for passengers that's far less jarring than the bounce that happens when the front suspension unloads when the car has stopped.
Fortunately it's easy to practice as stop signs are everywhere.
Nothing is more frustrating than coming to an intersection with full line of sight for miles in both directions, where you know there is no car that is close enough to hit you, and having to come to a full stop anyway.[1]
In my part of the US we have a lot of rotaries (round abouts). Thousands of cars go through them every day, no stopping and no accidents... it can be done.
[1] This is of course hyperbole. There are much more frustrating things.
Edit: In fact I thought of something better. Waiting 10 minutes at a red light when there is no one else on the road.
Four-way stop signs seem to be a desperate attempt to either avoid telling American drivers to "not be selfish" and allow other traffic through ahead of themselves, or to just avoid having to have drivers learn priority systems.
Avoiding major/minor road intersections seemed to be the main use of four-ways (around West-Chicago/Illinois, where I did most of my US driving), with ridiculous cases like four-way stop signs on roads where you could literally see hundreds of feet down each of the other roads, and not a car in sight except the cop car parked next to the intersection.
True stop signs in the UK are reasonably uncommon except in cases where they are actually required e.g. a junction could appear misleading.
Keep in mind that rotaries are extremely rare in the US, and that many four-way stops are replacing what would be rotaries in the UK.
Also, some intersections are more dangerous than they look. I remember a 4-way intersection with a 3-way stop that looked safe, but after the 10th traffic accident or so, the town said, "Let's just make it a 4-way stop," and the accidents stopped. One of entering roads was right after a major speed change (so people wouldn't slow down enough), and another had slightly worse visibility than it seemed. But once everybody was made to stop, the intersection worked safely, at the cost of adding 30 seconds of travel time.
Come to Massachusetts.
Others have been made in places (like a crevasse) where the circle is too small for a city bus to safely traverse - it has to jump the curb every time. The traffic light it replaced was actually slightly safer.
We can do that once we are at (or close to) 100% robot drivers. In the meantime, I'd prefer my robot drivers to err on the side of conservative.
> Humans violate the rules in a safe and principled way
that there are a lot of unwritten rules when it comes to road behaviour that traffic regulations does not fully capture.
In my experience, people always drive a bit faster than allowed. If you raise the limit, people raise their speed as well. Maybe the solution is to only allow the relaxed rules to robotic drivers? (though that would also raise resentment).
I think its more of people prefer to drive at the speed they're comfortable with (they believe its sufficiently safe given the environment), and legal speed limits tend to be significantly lower than that, except during special cases ie traffic jams. Im not sure if people are actually good at guaging where they should feel comfortable driving, but if you relax the rules for AI, it would be in a similar setup and the same question would apply.
But the thing is, that on the road people often drive not at the speed they're personally comfortable with, but closer to the average speed all drivers on a given road stretch are comfortable with. I.e. I've often been in cases where the driver drove faster than he/she liked because otherwise everyone else would stick to their tail, honk, or aggressively pass by.
major number of motorcycle accidents happen because people feel they have overview of whole situation and don't respect the simple traffic rules, both in cars and on bikes.
and yes, based on my 100k+ ridden mostly on european roads, following dead simple traffic rules of these days would prevent many accidents (and deaths). Maybe we would all get to destination 3 minutes late... so what (actually self driving should clean up the roads and especially traffic jams)
A lot of the traffic rules exist to remove the need for having total overview of the situation. That is, if you actually follow them, and can safely assume everyone else follows them too, then you don't need to see what's behind the corner and you can be confident you'll get to your destination safely.
I can imagine autonomous motorcycles. The hardware doesn't take that much extra space to make it infeasible. Whether anyone would like to use such a vehicle is another matter :).
Furthermore, I'd agree that increasing the limits above what humans currently already do would be a bad idea. At least as long as it is still legal to have human operated vehicles on the roads. However the issue we are discussing is whether you want AI to emulate some of the more impatient behaviours of humans. In many instances I think it would be safe. So it seems logical to bring AI up to our level of throughput.
Eventually the long term goal of driverless cars is reduced congestion due to the improved throughput provided by AI. But so long as there are humans also driving cars on the road the only way AI could improve throughput is to emulate human behaviour to some extent (ie not the worst of human behaviour but more our ability to judge when we need to navigate tighter spaces in congested junctions - for example).
- Assume that drivers who engage in the "more impatient behaviours" are increasing the amount of accidents compared to what would happen if they stick to the rules.
- Note that introduction of self-driving cars is a perfect opportunity to use them as a forcing function on the road, to make those impatient drivers stick to the rules.
- If both of the above is true, we can save more lives right now by having the self-driving cars stick to the rules than by having them emulate the impatient drivers - a task which the machines will handle fine, but which would keep such behaviour legitimized for humans.
Now I'm least sure about the first assumption; it intuitively seems true, but beyond occasional articles pointing out speeding and careless driving as primary sources of traffic accidents, I haven't spend time evaluating the data on that.
The second point seems intuitively true from experience of how drivers adapt to other drivers around them. Self-driving cars can be made stubborn by default, so everyone else would have no choice but to adapt to them.
Take learner drivers for example, people sometimes drive worse around them (pulling out in front of them etc) because they simply don't want to get caught behind them. This is even before they've sampled the competency of the learner. The L plate is often enough to make bad drives drive worse.
Thus the only definite way you can argue that AI would slow down bad drivers would be if the bad drivers were the ones being driven by the AI. In which case we are back to the point about whether AI can safely drive a little more aggressively (for want a better description) due to it's greater competency at driving.
If humans were required to report every bump and scrape, what would we find out? I suspect we would find that humans are just poor drivers overall, and "they don't drive like I do" is the number one cause of minor incidents worldwide.
My main submission to this argument is the ubiquitous rear-ending in the turning lane where the first car's driver starts to roll and then stops, while the second driver is turned to watch approaching traffic and thus oblivious to the behaviour of the car in front.
In contrast, I've found most speed limits in Germany to be reasonable absolute maximum limits - except on one or two places on streets I know very well. And I'm not talking about the Autobahn - if I see 70 km/h in Germany (45-ish mph), its probably dangerous to go faster. If I see 45 mph in the States, it seems to mean "everybody go 55 mph".
1. The traffic lights need to be coordinated and that means presuming a certain speed. Now the problem is a lot of traffic lights are close enough that changing the programming of just one should ripple through the entire system and this recently (especially on a regulation making time scale) was a problem consuming extraordinary amounts of computing power.
The other, more cynical reason for lowering speed limits is that in many countries now, speed traps are a significant source of income for the state. Making the speed limit significantly lower than what the standard 85% rule says is a good way to extract money from unsuspecting drivers.
Excessive speed is the primary cause of a third of all fatal accidents and a contributing factor in at least half. Instead of training the self-driving cars to be more human, we ought to train humans to act more like the computers! Speeding is one of the stupidest norms that exists in our society. We should definitely not try to encourage it.
https://en.wikipedia.org/wiki/Solomon_curve
> Speeding is one of the stupidest norms that exists in our society.
Politicians arbitrarily setting the speed limits is the root cause.
The EU has many countries which have higher freeway limits, yet lower accidents per mile driven. They also have more stringent licensing.
>Politicians arbitrarily setting the speed limits is the root cause.
Of speeding, or accidents?
The link sourced multiple studies. The originals were dubious, subsequent ones not so much.
But the issue the Solomon curve and other similar studies points out is that humans (and largely Americans) will ignore the rules if they feel they can safely go faster.
And sure, you can't ignore overall average speed. Crashes on roads where the average speed is 100 mph are going to be much more prone to fatalities than 55 mph roads. So, there are two rough equations: Lower average speeds mean less severe crashes; speed limits set to the 85th percentile mean fewer crashes. As the speed limit is raised (from 0), there will be at least one solution that satisfies both of those equations and results in the fewest fatal crashes (if that's your goal).
In certain situations the lower speed limits can make things less safe. For example, a straight, well maintained toll road with wide lanes, clear vision ahead, and say wildlife barriers that was designed for 75mph max speeds is constructed in a state with statewide 55mph speed limits. Construction uses the latest traffic easing and safety techniques, since you don't want to construct all roads to enforce speed limits that may change in the future.
So now most truckers and cautious drivers go 55mph, while people with less inhibitions are weaving in and out of them.
It doesn't mean increasing limits to something that is unsafe.
Counterpoint: creating traffic by going at the speed limit requires that other people on the road are going over. Both parties are to blame, but only one party isn't increasing their energy at a exponential rate (1/2mv^2).
On the flip side (almost literally), I once drove around Ireland for a week. Of course my instinct was to drive slightly over the speed limit, but I quickly found 100 kph on 1.5-lane roads winding around sheep pastures with no visibility nearly impossible to safely match let alone exceed…
This isn't obvious at all.
If two actors, one of which is following the law, has a higher risk than two actors both breaking the law, the lawbreaker is still at fault for the increased risk.
While crashing into an AI driver I may be able to blame the AI, it's still a huge pain and possible loss for me as a human.
If the reality is that AI is not leaving and they have peccadilloes, then the human drivers and legal structure will just need to adapt (something we aren't bad at).
Erratic behavior causes accidents, be it driving too slow, too fast, weaving, impaired, etc...
Now, if all the cars drove the same way, we'd be fine. We'd have very few accidents. However, we don't. So, we bump into stuff.
We don't actually tend to follow all the rules of the road. Like all changes in traffic, we will adapt. It takes time. Something like a rotary can take 18 months before traffic fully adapts. I can't speculate as to how long this will take.
I don't understand why we are not investing all of the efforts and resources that are now spent to develop full self-driving capibilities, into making existing viable safety technology cheaper, and eventually mandatory. I'm 100% convinced it would have a much bigger impact on traffic safety, in a much shorter timeframe, to the point we don't even need self-driving cars anymore. At least not for safety purposes.
Come to think of it, maybe the whole self-driving car hype is not actually because companies want to improve car safety, but because they see new business models based around them, and we, the public, need to be convinced self-driving cars will be safer than alternative technologies or modes of transportation.
And for some reason the minimal front distance for WV's ACC is almost an order of magnitude larger than safe braking distance, so other drivers see no issue with changing lanes into that large free space in front of you.
So in all, this technology obviously requires some work both in terms of reliability and in terms of UX.
I'm sure self-driving behavior is a factor here, but I also suspect that breaking and passing at intersections is just a common cause of zero-damage accidents. After all, one of the collisions in the story was a bike rear-ending a car - how many drivers would report that if there was no damage?
Anecdotally, it depends on how far away you see it coming. I'm constantly scanning front, rear mirrors, and side mirrors and if someone is driving erratically I take steps to get out of their way (in a safe and predictable manner of course!) I have no evidence this keeps me safer, but this and other "traffic calming" behavior I engage in makes my commute less stressful. It has the nice side-effect of continuous practice that reinforces good driving techniques.
.. and by "brace yourself", you would touch your head to your headrest so that your neck has nowhere to move/flex with the impact. You want all of your body to be completely rearward in your seat.
Also, while this may be morally negative, if there is another car in front of you, you could creep forward and lightly touch their car with your own, making yourself the middle ball in the Newtons Cradle[1] that your three cars now describe.
[1] https://en.wikipedia.org/wiki/Newton%27s_cradle
1. Why don't people pay attention to break lights? It's hard for me to relate to people who are always stopping last minute.
2. When someone is driving aggressively or erratically behind me, I break-check them as often as a I feel I need. Won't solve all rear-end issues, but it improves your chances of avoiding accidents.
This seems like an awfully good point. One of the incidents described was a bicycle rear-ending a car - reporting from that incident reveals a scraped knee, a scratched fender, and an intact bike. How many times a day do equally minor accidents between people get ignored?
Maybe self-driving cars really do get rear-ended more, but I haven't seen anyone try to account for reporting gaps like that.
http://www.thedrive.com/tech/11407/a-self-driving-chevy-bolt...
This is fine. There are no CA DMV reports of "autonomous vehicle entered blind intersection and was hit by cross traffic". Near term solution: put a rear bumper on your autonomous vehicle which meets the old 5MPH bumper standard from the 1980s. Longer term solution: manually driven cars need basic anti-collision radar, which is now standard on many high-end cars and all Volvos, and may be mandated in the US just as ABS is now.
Useful temporary fix: brake lights that go up to super-bright ambulance level when the autonomous vehicle's sensors indicate something close behind and there's a need to brake.
[1] https://news.ycombinator.com/item?id=15437322
this seems like a pretty clever solution if it would actually encourage the correct behaviour in human drivers. unexpectedly bright tail-lights might just have the opposite effect though, and further distract the driver who is about to rear end you.
We controlled all big highways.
The result: Over 95% of the cars were going way faster than what the rules were and nothing wrong happens. That is, rules are wrong because politicians decide them and they are always extra cautious, so they are not responsible for accidents, and they make money out of fines.
I see a big problem if everybody follows the law every time, like robots do, because rules are so wrong in the first place.
One of the big things with robots is that you could have real feedback with real data measuring in the real world. You don't need to guess, just test in the real world and measure the consequences for example, of increasing the velocity.
You could also change the parameters on real time when you know there is ice or snow in the road... you don't need to have the same parameters when it is sunny in summer and in winter.
I'm not sure what do you mean, there are plenty of accidents caused by speeding both on highways and within city limits. Care to elaborate how you conducted the study and/or link to it? As far as I'm aware extra caution is built in exactly to account for human errors and disregard for laws and have a margin of safety.
I can't help but feel maybe your professor missed an opportunity to teach about sampling errors.
A road can go decades without a crash, reduced speed will reduce severity of a crash when it happens.
Reduced speed usually increases fuel efficiency, which is also a factor in pollution control; in cities in particular that can have a big impact on health.
I would have thought speed limits are decided by technicians, based on statistics. I don't think it's elected politicians who decide them.
So when you break the rules, what principles (i.e. rules) are you following?
I don't share this optimism. I've been driven by folks who violate the rules in totally unsafe and scary ways (though "principled" in a way, in that they do it consistently -- unless a cop is looking).
The rules mean nothing, everyone drives freestyle. In towns and cities, it works extremely well because everyone is very attentive and ready for anyone else to do anything, at any time.
My experience in over 60,000 miles in those parts of the world through more than 30 countries would say they are an order of magnitude less crashes there than in US/Australia/Europe, in the towns and cities.
(Note: highway driving is a whole different story)
There's some truth to that, but there's also the ugly reality: statistically those places tend to have extremely high death rates (which may or may not come with high accident rates). Because while people are definitely more aware and cautious on the road in general, nobody will ever take the extremely dangerous drivers off the road, and they're now emboldened by the lack of consequences (both from the law and from drivers who mostly willingly give up their right of way to whomever is most reckless).
I know someone whose entire family was killed by a driver who hit them head on due to blindly passing a truck while going 160kph on a 40kph road in Honduras. I know for a fact that his wife was an extremely cautious driver, but you can't avoid something like that no matter how hard you try to predict the unpredictable.
In the cities, freestyle driving works great, and I would bet the accident rate is much lower than "rules based" driving.
On the highways is a totally different game. I have seen some horrific accidents, luckily I have avoided them all so far.
Well if that is the case, then hopefully they will be ready for self driving cars to actually follow the law.
Do you have any statistics that indicate that there are 90% fewer accidents in Latin American and West African cities than in Western, adjusted for the number of drivers? Seems rather optimistic.
Certainly not scientific, though other people I know that have driven the world agree, driving "freestlye" is safer than following the rules.
The article suggests that, though self-driving cars don't crash into other cars, they "cause" crashes (mostly fender-benders) by driving in a different manner than humans. No evidence that they cause more fender-benders than humans, who drive all sorts of different ways. Apparently, the author thinks this is so obvious, no data is required.
Then, the article seems to propose that self-driving cars should be like humans, who allegedly violate laws in a "safe and principled" way. This ridiculous statement ignores the tens of thousands of people killed in crashes caused by human-driven cars every year, and the many tens of thousands severely injured, by these "safe and principled" violation of driving laws. Most accidents do not involve complying with traffic laws, actually. That's why we have traffic laws -- to prevent this carnage. Individually, however, we still tend to believe that even if we engage in risky behavior like speeding or running lights, we're not the one personally who will crash.
As to some people who think self-driving cars should speed because so many people speed, please check the mortality rates for accidents at various speeds (they're widely published) -- survival rates go down lots as speed goes up, especially if a pedestrian is involved. On the other hand, if you do the math, on most trips the amount of time actually saved is much smaller than the exhilarating sense of going faster than the next guy you get from even moving a couple of mph faster. This is why many major cities are reducing speed limits and taking other steps to reduce traffic speeds. Speeding, the most common rule violation in driving, is neither safe nor reasonable, and the author, supposedly an expert on driving, cannot provide any evidence to indicate that it is. Well, except that he seems to assume that tens of thousands of traffic deaths are acceptable and unavoidable.
Perceptive readers will notice that I haven't cited stats to support my assertions in this comment. That's because it's a comment and I'm not holding myself out as a journalist who's done sufficient research to report on issues of motor vehicle safety. I'm a mere commenter, but be sure, if you do the work to find out what the real numbers are (regarding both driving safety and actual driving times) it bears out what I'm saying.
Thanks, it's nice to start my morning with a good laugh.
Maybe the author lives in some magical wonderland filled with good drivers, but this is not what I observe at all. People drift out of their lane for no reason. They careen past the stop line because they can't be bothered to push the brake pedal a bit farther down. They refrain from signaling because they just don't care. They blast through stop signs and red lights without heed to traffic or pedestrians. They brake spontaneously in the middle of a fast moving road. They drive fast in dangerous conditions and slow in safe conditions. They spend half their drive texting because they think they're the one person on the planet who is able to safely text and drive. How is any of that safe or principled?
Try driving in Asia and get back to me.
(Disclaimer: I'm hard line traffic law follower.)
One of the things you learn living in France is a sense of the rules that make no sense and that everybody is probably disregarding, and the ones that are actually useful and probably enforced.
Always smile to a clerk. They will tell you which mandatory paper is not mandatory and which rule you can break. They will also tell you which rule the state is incapable of following and how to get around it.
I live in Paris, and although I do not drive I have the impression that most drivers here are extremely inconsiderate. As an example, there is a crossing which is constantly congested with buses and lorries stuck in the middle here. This is mainly because nobody respects the rules or priorities. When an officer comes in to control the traffic instead of the lights, it gets "magically" cleaned up and the circulation is fluid.
You are actually providing an example that the Parisian way of doing things is not as benignly efficient as you suggested in your first post.
The fact that you can not change lane without forcing is a normal consequence of the increase of the traffic. I think the unwritten rule is that people should not turn on flashing lights too early before acting. The better communication of the intents and a fair level of trust that other drivers will understand it (and not try to block you) makes driving a lot easier (in Paris).
A bunch of highly nuanced rules like this seems very fragile. If you are driving in this manner, you have a responsibility to be alert for things breaking.
The facts show that whatever the parisians are doing is definitely the wrong way to do it, and should be changed.
'We'll all just avoid making any mistakes' is a really unwise attitude anywhere, but particularly when driving a car.
There's no better realization of distributed peace. Who is the self-serving person who thinks they are entitled to travel to Paris and drive around without understanding the cultural norms of their unspoken driving rules?
Is it a reality? Sounds like conjecture to me, and the fact that many humans are killed in traffic fatalities certainly suggests otherwise.
Any other such commonly observed violation would probably have to be studied to answer the question - and given humans' tendencies to take on a fairly uniform level of risk no matter what the rules are, I suspect the question may need to be "does this keep the risk level constants while increasing throughput" rather than "does this reduce risk".
So ... it's plausible both ways so far as I can tell.
Who is causing the danger of driving the maximum legal speed.
Also, it's rare that all of the traffic flow is above the maximum legal speed. The ones speeding then cause the differential, which is the danger.
The fault is not with those not keeping up. But rather with those breaking the law because it's convenient.
The fundamental question is who is at fault for the differential in speed (i.e. that which increases risk of injury/death with an accident)?
The answer is the person who breaks the law and the one who does so as a precaution to themselves (i.e. following the flow of traffic). An accident by this group disproportionately affects those obeying the law.
So I see three groups.
1. The speeder. 2. The law abiding citizen. 3. The person who is aware of their surroundings and adapts.
Ethically between 1 and 2, 2 has the upper hand. 3, when driving with 1 increases risk to 2. But if 3 joins 2 they increase risk to themselves and to the larger group 1.
Absolutely. I went into some details here: https://news.ycombinator.com/item?id=15495904
You also have to weigh the lives against the value of the time saved -- the whole reason those norms exist. There's no way it comes out in favor of the deaths.
Let's try and ballpark whether it's worth it to speed 5km.
Roughly 410 billion miles driven yearly by the French.
The average speed of a journey in France is ballpark 60km/hr
Average salary is 36k in France with a 1450 work hours a year for €25/hr.
Putting it all together we get €0.42/km or €170B.
Now suppose that everyone always drives 5km over the limit bringing that average to roughly 65km/hr or €.38/km or €156B for a ballpark €14B in savings.
Say you can save 10% of those deaths by driving the speed limit which is extremely generous. I would take odds that not even 1% of the deaths could be traced to speeding 5km over but let's pretend. Even then each life saved still costs €400M.
Is it worth it?
But the intent is not always sufficiently appreciated.
Case in point, urban speed limits at 50Km/h in EU or 35MPH in US. For most drivers (me included), they feel terribly slow and overly precautious. Yet the intent has to do with stopping distance. In those areas, you should be prepared to come to a full stop on a very short notice because a child/dog/ball could be darting out in front of you with zero notice. Not only that but the cars behind you should also be able to stop. Whenever you're driving at 45 in a 35 zone, keep that in mind: not only you need to be ready to slam on your breaks but you must have enough distance with the vehicle in front of you, should they need to slam on their brakes.
Most people only consider the "happy path" scenario, not the "error handling" one.
Let me take a thougher position. The value society gains by upping the speed limit on a major road will quickly outpace the value of the lives lost due to accidents. It's just that the injuries and deaths are more visible.
The evidence should be that almost everyone drives at the faster speed anyway and 35mph roads aren't death traps.
This is self evidently not true though. Humans are terrible drivers, and the road would be safer if all of those people who are unsafely not following the rules instead followed them.
The reality of the world that we live in is one of terrible drivers that get into lots of accidents.
Exactly. This is why the speed limit on highways is always min("one or two under the misdemeanor speed in this state" "as fast as conditions allow").
This is the part humans (who almost invariably overrate their abilities at everything, especially driving) are really bad at determining.
Anyone licensed to drive in France, actually. The requirement is that they understand the traffic laws and regulations.
Not so simple.
The rules are created via legislation affected by lobbying.
The AAA for decades have been lobbying to keep speed limits low.
In the perfect world, speed limits would be informed by the 85% percentile speed of engineering surveys.
In reality, it's much different.
If a large number of drivers are driving at a speed higher than a prima facie speed limit, the speed limit could actually be low, in which case, the drivers actually following the prima facie speed limit are creating a hazard to the flow of traffic.
Driving on highways is easy, and solving that is still valuable, but it's going to be amazing to see these things work in situations like those.
:(
https://news.ycombinator.com/newsguidelines.html
Second, we're not going to replace all cars with autonomous cars overnight, so autonomous cars need to be able to deal with unpredictable human drivers, even if only temporarily.
Thirdly, in my experience, it is oftentimes the overly cautious driver that causes more accidents than anyone else. If the self-driving cars are emulating the people who do the exact speed limit on the highway in the left lane, this is not a good thing because the deviation from the established norms of any given highway is the best indicator for where accidents are going to occur.
In my driver's ed course, it was explicitly covered that police will often not pull you over for doing 5 mph over the limit. If we're training drivers to casually break the law, then it's only logical we'd be doing the same thing to self-driving car AI.
It's really weird that you use driving the speed limit in the left lane as an example of dangerous rule-abiding behavior. That's against the rules in most jurisdictions, and definitely should be in any where it's not.
I agree that it's possible to break the rules safely. Rolling a stop sign in a situation with good visibility or exceeding the speed limit by a bit is perfectly fine. I agree that humans sometimes break the rules safely. But I disagree with the statement that humans violate the rules in a safe and principled way. When humans break the rules, it's almost never based on any principle besides sloth or selfishness, and while humans do break the rules safely sometimes, it's almost always safe by accident.
The larger point that autonomous vehicles need to not be slaves to the rules is a good one. But humans are not the model to emulate!
You weren't quite that clear about it in the first post, the rolling stop sign with good visibility was my first mental picture while reading, so I apologize for the misunderstanding. That being said, all of these maneuvers you're talking about could easily cause an accident whether the other vehicle was AI driven or not, so I don't see how it's relevant to the topic at hand: that corner cuts and slights which are technically violations but not inherently unsafe need to be predicted and handled by the self-driving car.
> It's really weird that you use driving the speed limit in the left lane as an example of dangerous rule-abiding behavior. That's against the rules in most jurisdictions, and definitely should be in any where it's not.
Again this was my mental picture, the idea of a robot driving according to the letter of the law reminds me very much of the sort of people who cruise at speed limit in the passing lane, just for different reasons.
> I agree that it's possible to break the rules safely. Rolling a stop sign in a situation with good visibility or exceeding the speed limit by a bit is perfectly fine. I agree that humans sometimes break the rules safely.
Agree.
> But I disagree with the statement that humans violate the rules in a safe and principled way. When humans break the rules, it's almost never based on any principle besides sloth or selfishness, and while humans do break the rules safely sometimes, it's almost always safe by accident.
I would assert that the times they break them safely is not always safe by accident. Every given decision in driving is a balance of risk and reward, the reward more often than not arriving at one's destination faster, and that's where the majority good decisions come from. The bad ones are often not decisions of R&R but of distractions, as you point out like texting, or worse, inebriation. I'm not arguing human drivers are better: I'm merely trying to say that the article is, I feel, attempting to demonize people for driving like people drive, and saying self-driving tech needs to be able to handle the unpredictable nature of human drivers, just as I did on my commute earlier today, if it's going to be a mainstream technology.
Of course autonomous cars need to predict and handle sloppy human driving. But that's not what the article is talking about. It's talking about programming the cars to break the rules "in a safe and principled way" the way humans do because humans will expect it. I don't even have a problem with programming the cars to break the rules in a safe and principled way. Where I have a problem is when the article points to humans as a model because that's what they do. It's not.
> Every given decision in driving is a balance of risk and reward....
This is probably the center of our disagreement. Every given decision should be a balance of risk and reward, yes. In reality, for 99% of drivers, every given decision is a balance of reflex, selfishness, and vague remembrance of rules they learned 30 years ago and forgot 29 years ago.
It's true that safely breaking the rules is not always an accident. It's merely an accident 99% of the time.
There's a busy intersection near my house which I cross on foot fairly often. Cars routinely turn right on red without stopping first. Every time I cross that street, I have to carefully watch the oncoming cars until it looks like they're going to stop, because if I just start walking when the walk sign comes on, I will get hit. These drivers aren't risking my life because they're balancing risk and reward. They're doing it because they want to beat traffic from the other road and are too clueless to understand that what they're doing is dangerous.
Again, I have no issue with programming autonomous cars to accommodate flawed human drivers, or even with programming them to deliberately break the rules in cases where it's a net gain. I merely object to the proposal that humans are the model to follow in that respect.
I feel like this is going to be a huge problem if we have mixed human operated and self-driving vehicles on the road at the same time. If we ever move to entirely self driving cars we could move to some sort of protocol for cars to communicate about ambiguous self driving actions.
It's really common to see the Uber Volvo's driving around, but it definitely causes me to tense up when driving around them. I have no idea why, I realize there is significant risk when around any other car on the road. It's new. Hopefully these issues get worked out in a reasonable time frame, but it is software.
[1] https://en.wikipedia.org/wiki/Pittsburgh_left
The real issue is not self driving cars, it's poorly designed intersections.
It's easy to insult design but very rarely are intersections, especially in Pittsburgh, easy to design.
The bigger one is to make sure that the next traffic signal down the road is also green. To avoid traffic piling up on an intersection because it has nowhere to go. Of course, this should also depends on the amount of traffic.
Any city can in theory build systems where traffic that wants to go to empty roads gets more green and any traffic that wants to go to full roads gets less green.
https://xkcd.com/793/
Honestly, I'm amazed urban traffic is as good as it is, since the core problems are usually irreducible. (i.e. more people want to go to a place than fit there.) A quick look at gridlock in cities where traffic signals aren't carefully planned and widely respected shows just how bad things can get.
(Suburban traffic is another story, but that mostly seems to be about overbuilt roads with no actual planning at all. My hometown has a couple of intersections that literally don't function - you can get caught at a permanent red light.)
Err... I modeled traffic.
The possible solution would be roundabouts and yield signs, but that would be expensive, require redesign, and we humans pretty much suck at roundabouts.
So, these things are very much fixable given sufficient priority and resources.
Pittsburgh isn't flat, many times it is not "move this road." It's bore through this mountain, build this bridge, fix the drainage issues around this creek, etc. and fixing intersections almost always leads to more issues downstream.
Your comment also reminds me of a proverb to the tune of "bad drivers aren't a problem until they meet".
Sure, one can be that driver that leaves death and destruction in their wake but still look good on paper if the metric is legal and financial penalties since those only loosely correlate with actual bad behavior.
Why exactly would they follow the new rules?
I am curious how we sort this out. Will roads become stricter with cars self enforcing the limits? How do self driving cars know who is next at a four way stop?
There is not a hard law that states speed limit at least in California; instead the law sets 'reasonable and prudent' as the limit, and that lawfully set and posted or legislated limits set forth evidence of what is reasonable and prudent. Driving over 100 mph is separately off limits, however. (I would rarely find that speed to be reasonable or prudent!).
A self-driving vehicle would likely be recording evidence of conditions which could show that the speed was reasonable and prudent given the conditions, which include other traffic on the road.
The basic California speed law is:
> 22350. No person shall drive a vehicle upon a highway at a speed greater than is reasonable or prudent having due regard for weather, visibility, the traffic on, and the surface and width of, the highway, and in no event at a speed which endangers the safety of persons or property.
It is furthered by:
> 22351. (b) The speed of any vehicle upon a highway in excess of the prima facie speed limits in Section 22352 or established as authorized in this code is prima facie unlawful unless the defendant establishes by competent evidence that the speed in excess of said limits did not constitute a violation of the basic speed law at the time, place and under the conditions then existing.
Additionally section 22348 describes a specific additional infraction for driving at a speed greater than 100 miles per hour.
From an implementation perspective that sounds like a lot of overhead just to let Californians drive fast.
So it's not a 49/1 rule.
And, in any case, 100% of US jurisdictions prohibit driving at a speed above the safe and reasonable speed if that is lower than the posted and/or statutory limit, so you need to be able to determine what is safe for conditions anyway; it's just a minority of states where that is a relevant determination when the results are above the posted limit.
> There is not a hard law that states speed limit at least in California
Incorrect.
I don't know what you mean by "a hard law", but, in California, there is a section, from the very statues you present, that sets a statutory maximum limit.
The complication, being that there are three maximum speed limits in California: 55mph, 65mph and 70mph, depending on the scenario.
CVC §22349 sets a maximum speed limit of 65 mph throughout the state (except for the 70 mph limit on some freeways).
CVC§22349(a) reads: Except as provided in Section 22356 [70 mph on some freeways], no person shall drive a vehicle upon a highway at a speed greater than 65 miles per hour.
The CVC§ 22356 referred to above allows a 70 mph speed limit if justified by an engineering and traffic survey where a 65 mph limit would otherwise be applicable.
Try arguing the 22350 while you have been cited the §22349 on a ticket. Not going to fly regardless of what one might read on websites online.
I mean, you're obviously correct, human drivers do idiotic and deadly things all the time. But I think we can distinguish that from the behavior in the article.
Gently rolling a stop sign at a high-visibility intersection sounds like a safe and principled violation. Driving around a car waiting to turn left is a bit less safe, but still a principled or predictable violation. The Pittsburgh left is a (fairly) safe and principled violation.
None of those things are comparable to texting and driving, or ignoring a red light. They're informal adjustments to the traffic code that other drivers really do anticipate, and which aren't nearly as dangerous as general "bad driving".
Yes, there are people who carefully evaluate the situation and only roll their stops when it's safe. They make up a minuscule percentage of the population, though. Most people just do it out of laziness or selfishness, and when they do it safely, it's a happy accident, not a deliberate policy.
The problem is that humans easily misjudge what they believe to be an empty, clearly visible intersection.
The lesson: just because it's your turn to go does not mean that others will not enter the intersection when it's not their turn. I agree that rolling stops at a 4-way stop are generally safe when you can clearly see that no other cars are at or approaching the intersection. But in my experience, the fact that the other person is at fault has been little consolation for having to deal with a totaled car and a permanent backache.
At the end of the day you either trust other drivers or you don't. You're not exposing yourself to any significant addional risk by making these 'safe violations' than you do driving in general.
Who died defending his right-of-way.
He was right--dead-right--as he sailed along
But he's just as dead, as if he'd been wrong.
I just ran the math, the kinetic energy of a 2000kg car at 5mph is the equivalent of the kinetic energy of a 100kg bicycle+cyclist at 22mph.
I absolutely roll stops, but it's important to recognize what we might miss. My old commute had a series of 5 (virtually useless) stop signs in a row, an arterial with very little cross traffic but tons of 4-way stops. (should have been roundabouts). The routine of the drive, and the necessity to stop, the focus on the stop signs and stop lines, and the almost total lack of pedestrians meant that I'd occasionally start off before noticing the cross traffic or pedestrian. (yes, this is more of a boy-who-cried wolf kind of thing). Also, modern cars are amazingly difficult to see out of-- pedestrians can be hidden behind thick airbag-containing A pillars. When you're moving, however slowly, everything in your field of view is also moving. When you stop, it's much easier to spot objects that are NOT stopped.
Hell, just last week I was making a left turn on my bicycle. As I approached the intersection, I slowed, looked each way up the road several times as my sightlines improved, and just at the moment I committed to my turn and began to proceed, I saw the minivan coming from my left that was on a collision course with me; I locked up my brakes and stopped just in time. I know I looked exactly where the van was coming from, multiple times, but somehow missed it. I suspect, again, our brains are much better at seeing what's in motion when we are ourselves stopped.
I have been running for 25+ years and have on occasion been traveling at close to 10 mph and usually at least 8 mph. So California rolls are pretty much the bane of my existence.
This is primarily because the "roller" may be planning to take a right turn and are looking straight and glancing left and don't even look my way traveling from their right.
I do take responsibility for my own survival so I wear lights, and try my best to make eye contact before entering intersections etc. but I have had many close calls where they did not even see me due to their own target fixation.
Now that I am on a bike more often I feel even more insecure because my speed is so much higher.
If there are any pedestrians around then you don't.
It looks a lot more insane when you're paid well enough to outsource those behaviors (and liability) to whoever's driving your Uber.
It's like when a company makes a statement about supporting a living wage and contracts out its janitorial services who pay anything but.
You do not even notice the small ones, which happen all the time. How many people cross solid lines? How many people sometimes drift a bit out of their lane? How many people scrupulously follow the speed limit like a computer does, not even accidentally drifting above it? How many people simply screw up the order of operations at a busy stop sign? How many people always scrupulously use their turn signals in accordance with the local laws? (You may use them frequently, even scrupulously, but I bet you often turn them on what is legally too late.) I've been driving through a construction zone all summer. People definitely don't drive "45 when workers are around". They don't slow down when the workers are behind the concrete barriers and a dozen feet from the road, and may drive 25 when they're sitting right next to the road with just a barrel a bit between you and them.
Even a cop tailing you and desperately hunting for reasons to pull you over still probably lets through plenty of violations. There's too many laws for the cop to keep in their head.
Meanwhile, a lot of highway on-ramps are painted with single-lane-width, high-angle merges that are dangerous to adhere to - they leave you making a ~30 degree turn while accelerating into traffic. As a result, everyone starts their merge early in the gore area, matches speed, and merges at a gentle angle.
(The road designers presumably know this, and keep the marked path narrow regardless so that drivers won't attempt to form multiple lanes.)
This is a pretty great example of what I'm describing - I don't think people make great decisions about violating driving rules, but I also don't think it's fair to call safe decisions "inadvertent". Not breaking that law is genuinely dangerous, and people violate it specifically to be safer.
The CBC had some nice coverage of this.
Allow me to fill in between the lines:
"What is this, some sort of (mostly still) free country where people are not given harsh punishments for all behavior that prioritizes the individual over the group?"
(sarcasm implied)
"By accepting a modest delay in going straight, the opposing traffic has saved the left-turning traffic waiting an entire light cycle to turn left, as well as saved an equivalent amount of time for all the cars that otherwise would have been stuck behind the left-turning car. In situations where there is so much oncoming, straight traffic that a left turn would not be otherwise possible during any part of the light cycle, the Pittsburgh left can allow a line of left turning traffic to proceed incrementally."
https://en.wikipedia.org/wiki/Pittsburgh_left
If people just wanted to turn left, this would happen everywhere. Instead, I see it almost exclusively in cities with narrow streets lacking left turn lanes. In extreme cases, it's even personally beneficial for the yielding driver - the alternative is sometimes gridlock that circles back around to them.
None of which makes it safe, but it's pretty obviously common for reasons other than selfishness.
Thank you. I'm going to use this reasoning next time someone complains.
(We don't seem to have this problem out here in California, where every street has at least four lanes!)
Wikipedia does hint at a legal kludge, though - "I judged oncoming traffic to be moving so slowly that I could safely turn". The idea I guess being that it's not a change to driving rules, it's just slow acceleration and a legal turn.
I can't imagine that would hold up real well legally, especially if people are flashing headlights and making an invitation. On the other hand, I've never heard of this leading to a ticket in Boston - it just can't compete with the other violations that happen constantly!
If you do this yourself, please reconsider. My experience is that drivers who do this (especially at 2-way stops) are watching carefully for perpendicular traffic while blithely ignoring pedestrians approaching the crosswalk that they are rolling through. As a pedestrian, I consider this the second highest threat to my personal life+limb, after left turners who (again) watch for cars and not pedestrians.
If you introduce the danger then you should take responsibility for safety.
To give one example, drivers are required to yield to pedestrians in cross walks and to pedestrians approaching crosswalks while pedestrians are only required to "exercise due care," which is a much weaker requirement. If you want to see how weak this requirement really is, look at the case law. In many cases in which the average driver thinks the pedestrian is in the wrong, the law disagrees and would place the lion's share of blame on the driver.
Is that actually the case? I know that the former is true, but I haven't come across a law stating that drivers are required to yield to pedestrians who are not already in a crosswalk. Does case law differ from the text of the law it refers to?
> while pedestrians are only required to "exercise due care," which is a much weaker requirement. If you want to see how weak this requirement really is, look at the case law. In many cases in which the average driver thinks the pedestrian is in the wrong, the law disagrees and would place the lion's share of blame on the driver.
I would be curious about what case law would say about a collision like this: https://www.youtube.com/watch?v=io_XBK0twtQ. Specifically, who would be found at fault/liable.
Heavier, faster-moving vehicles are typically expected to yield to slower ones, whether on the roads or on the local multi-use path, because they're the ones likely to cause the most damage when a collision occurs.
I did this when I lived in a rural setting and there were no sidewalks or crosswalks. In that case, I'm talking about rolling stop signs at entirely-empty intersections, where you can see that there are no cars or people around. At that point, I'm not sure I feel so bad about it - seeing a person at all is an adequate cue to drive differently.
I don't anymore, though, because I live in a city. Noticing everyone and working out their intentions means actually stopping.
Which, I suppose, is the sort of difference we might get between a self-driving car and a responsible human driver; the car isn't likely to treat stop signs contextually.
It's true that a self-driving car will likely not be designed to substitute it's judgement for those of the authorities placing signage, though I'm not convinced that this is a bad thing for safety, or that the predilection of some human drivers to disregard such signage is a good, or even harmless, thing.
A lot of times, they will be placed as a form of speed control. The MUTCD has requirements regarding the use of stop signs, but they don't appear to be followed most of the time. Otherwise, we wouldn't have nearly as many stop signs at intersections where there is plenty of visibility of cross traffic.
Of the two parts pertaining to complying with a stop sign (first stop, then yield), the second is far more important than the first when it comes to avoiding a collision. You can certainly come to a full stop and subsequently fail to yield and then collide with a pedestrian or vehicle. But even if you don't come to a full stop and do yield to a pedestrian or vehicle that has the right-of-way, you're not going to collide with them.
An interesting thought: what's the lowest speed limit sign you've ever seen?
I've seen 5 MPH in parking lots, but never on an actual road. For official road signs, I virtually never see anything below 15 MPH. I'd be pretty surprised if this was actually the ideal driving outcome. There has to be some intersection at which 5 MPH and a yield sign is the ideal arrangement.
But given how traffic laws are enforced, and how people tend to ignore low speed limits, it's not a very practical arrangement. Instead, stop signs go in everywhere that very low speed limits are required.
(I'm still vaguely wondering whether a lot of people in this thread have ever driven somewhere very rural. Gently rolling a stop sign at an intersection with a mile of visibility and no one else around is not some wild risk.)
http://www.wheels.ca/news/auto-know-jumping-light-for-a-left...
Although I'm in favour of the law in general, I really dislike the 'stunt driving' name. I'm not sure that many (or any) of the offences covered under the law really qualify as stunts. Dumb and dangerous, yes, but not stunts.
Call this law something else, and save 'stunt drving' for people who drive their cars on two wheels and other related craziness. :)
Yes, people do drive differently in different regions. Boston is the only place where I have had two cars totaled. It was parked both times. The second time the driver of the other car seemingly backed up and smashed into it again.
I love that dirty water.
I've always heard this called a "California stop." Are you in California? Looking for a point of data.
http://www.motherjones.com/kevin-drum/2015/10/was-california...
Google will find more results.
I'm coming up to an intersection and I see someone edging out, so I slow down and prepare to panic stop in case they pull out in front of me. They don't, so I proceed. That's good defensive driving, but a failed prediction.
I see someone with their signal on in the next lane, so I back up and give them some room to move over. Turns out they just forgot their signal was on. Again, good defensive driving, but an incorrect prediction.
People don't break the rules in anything like a consistently predictable manner. Not around here, anyway.
Noticing some behavior and thinking of all the possibilities is not predictive behavior. What you are doing is making a reaction based on the action by another driver.
In one case, you notice a driver edging forward, so you slow down in case they pull down. If you were predicting that they were going to pull out, you wouldn't just slow down, you would slam on your brakes.
In another case, you notice a driver with a blinker on, so you assume they are changing lanes, and slow down to give them additional space. This one is closer to a prediction, but in reality a driver would notice the blinker before they get anywhere near the car in question and would have plenty of time to tell whether it's a forgotten blinker or a conscious attempt at changing lanes. Or they don't see it from far away because traffic is moving very fast and/or is very heavy, so they just keep going making the prediction that the car w/ a blinker on will wait until the lane is clear (aka after you pass them) to change lanes.
The author of the article has not said those words. They were said by Karl Iagnemma who is a scientist from MIT, and the CEO of a company that makes software for self-driving cars.
It didn't sound like something a reasonable person would say so I tried to track down the context of the quote, and it seems to be a mystery. My guess is that the quote came out of a larger interview, and it obviously meant to highlight the fact that 'standing out' by following rules in a way that other drivers are not expecting you to, makes things much more worse. That makes sense to me, as opposed to your interpretation.
I can see a few interesting govt things coming out of "Autos":
1. Revenue from bad laws drops to 0. These vehicles follow the letter of the law. And they have the logs to prove any sort of claim against the former.
2. A false claim is filed (BLM makes about a point about this..) and says driver did X. Is this the person's responsibility, or the company who wrote the AI?
Humans contribute to lots of accidents when they unconventionally yield the right of way. Other drivers which should be yielding are now faced with the unpredictability of the too-cautious ("dangerously" cautious IMO) driver re-claiming their right of way. Or they face the problem of other drivers in other lanes who don't notice that the incorrectly-yielding driver is letting another vehicle into their right-of-way.
At sea the COLREGS require ships with right of way to maintain their course and speed (unless of course they have to give way to another ship, or make evasive maneuvers). Of course fishermen tend to ignore this rule, changing their course and speed whenever convenient to them.
Uh, yeah, I've had too many close calls with people merging to just blindly assert my right of way every time.
The worst-case scenario when yielding properly, with unconventionally given right-of-way is that both parties are at a complete stop and have angry drivers behind.
Now, if you are disobeying the law and are following too closely and collide with someone in front of you, that's a different situation......
The interesting tech is not only AV but the infrastructure where cars communicate automatically about which lanes there in, what their intentions are etc.
The current use case I think is cool is a car comes up to a slow truck and the truck communicates to the car that the passing lane is clear.
I don't think that this will really work. Considering how interaction between tech of different manufacturers works in other areas, I doubt that cars will be able to safely communicate within milliseconds in high risk situations. But they won't need to, they can just use their own sensors to detect what happens around them.
"One consideration in implementing the ship's autonomous functions was making its movements seem human-like. Any maneuvers it makes to avoid collisions have to be detectable by crew on other boats, so changes in direction (which are more easily spotted) are preferable to changes in speed that might be just as effective but are less obvious signals."
Obeying rules is nice but you need to account for behaviour of human actors as well.
People will rapidly learn to recognize those "damn robot drivers" and adapt the same way they do now when they see the "student driver" placard or the unmistakable sway and surge of an intoxicated driver.
Unfortunately for all those with fantastical fantasies about our autonomous car future, when they finally get here they’re going to be slow. And that’s a good thing, even if it means the utopia is still slow and congested.
Drivers expect other drivers to follow the rules of the road, not to be nice.
Did you hear about the woman in Canada [1] who tried to "be nice" and stopped in the middle of a highway to let some ducks cross? She killed a motorcyclist and his passenger.
Don't be nice -- be predictable.
Same is the reason why the bots get rear-ended -- everyone expects them to roll through a stop, who in the world does a complete, 3 second stop?
[1] https://www.google.ca/search?q=Quebec+woman+stopped+for+duck...
Here are the rules by state in the US.
http://www.ncsl.org/research/transportation/pedestrian-cross...
You should probably learn the rules of the road before you kill someone.
And rolling a stop is not following the rules of the road, it is a dangerous violation.
I don't think that's true on a highway or similar road. Stopping in the middle of the highway is very dangerous. You cannot blame the 150km/h+ cars to be able to stop safely in all situations.
It's definitely true on a highway (which is anything that's not an alley), and even a freeway (which is probably what you meant.)
> Stopping in the middle of the highway is very dangerous.
Failure to maintain adequate following distance on any road is dangerous; encountering stopped traffic or needing to stop rapidly, even on freeways, is a regular occurrence.
> You cannot blame the 150km/h+ cars to be able to stop safely in all situations.
I don't blame cars for anything, I do blame drivers for failing to maintain a safe combination of speed and separation.
And there's virtually no road, freeway or otherwise, where 150km/h+ (93mph+) is a safe and reasonable speed for any driver, irrespective of separation, so I can quite easiky blame a driver for choosing to drive at such a speed in the first place.
You will absolutely be at fault and stuck with the bill in every single rear-ending case, good luck arguing otherwise.
https://beta.theglobeandmail.com/news/national/quebec-woman-...
She was sentenced in December 2014 to 90 days in jail to be served on weekends, three years' probation and 240 hours of community service, and given a 10-year driving ban.
Regardless, it was dumb of her to stop there. But nowhere near as dumb as your idea that crosswalks are marked by definition. If you're going to give driving advice, don't be wrong. Especially on something like that, where you could easily kill someone and where the penalty could be as severe as manslaughter.
For example: corner to corner crossing is a normal law for pedestrians. Technically, you should always yield the right of way for pedestrians to cross in this manner. Say you are on a two-lane main road through town, and there is tons of traffic. Do you stop at every single corner to let one person cross?
The law says yes. Everyone waiting on you is going to say HELL NO! The people that stop at every corner think they are god's angel doing the right thing, but really they are being a massive sh*t head.
That's just one example of the fuzzy line between breaking the law for the common good or following the law to a tee to the detriment of everyone on the road.
Obviously, the best solution is just better city planning so that pedestrians rarely have to interact with traffic. When they do have to interact with traffic, it should be limited and explicit ( red lights, big marked crosswalks, etc).
Except your “common good” isn’t common. It benefits drivers by mere seconds in exchange for a mortal risk to pedestrians.
Like it or not, but the law is written as is for a reason: pedestrians are the lowest common denominator of society. Everybody is a pedestrian some of the time. When your expectations are for pedestrians to go several minutes out of their way to find a light and crosswalk so they can cross the street, just so you don’t have to slow down for a split second, it is you that is being the complete shithead.
Luckily autonomous cars will soon make all of the enpathy-deficient sociopathic shitheads wait their turn, like they should have been doing all along.
I don't think pedestrians should go way out of their way to cross, I think they should just wait for gaps in traffic.
Likewise drivers should use good judgment and stop for an old lady, but keep going if it's an in-shape jogger.
Maybe the robot drivers will finally drive home how dumb some of our road infrastructure is.
I shiver at the thought of having to cross any of these if there was only a yield sign. The principle that I have priority is not important enough over not getting hit by a bad driver.
TL;DR: I agree 100% with your point, but I do think bad drivers should be taken into consideration.
There seems to be a growing use of traffic circles (with yield signs) in the US, which are very much like playing real life Frogger as a pedestrian: cars coming quickly around the circle, often not 100% comfortable with how it works, watching for cars around them instead of paying attention to people crossing at the exits. In a busy intersection you sort of have to step out and try to get someone to stop for you or dash across a gap in traffic–a firm stop required for cars makes for a much more pleasant crossing.
There are stop signs that shouldn't exist but they're the minority.
I would like to know to what extent are accidents caused by robot drivers being "odd" versus humans not paying attention and then blaming the robot for being odd.
(ED: grammar)
However, when it comes to something that threatens these industries...it's suddenly the humans that need to lighten up?
I'm ready to get rid of the entire auto industry once and for all, along with all of the insane amount of income they generate off the backs of regular people.
I recently passed my California driving test, and I’ve observed that the placement of STOP signs here does not correlate with safety in the same way as Australia and lots of people roll through them.
I think self-driving cars would do a lot better in Australia.