> Similarly, unmanned vehicles might proceed at speed through an intersection where a stop sign has been removed by college students or knocked down the night before by an impaired human.
But so would human drivers. In fact, a self-driving car may actually do better in this scenario because it would have metadata stating that the last 3,000 times a car encountered this intersection, there was a stop sign here.
I don't think it does. Sure, a car would know that a stop sign was there the last time it drove through that intersection, but you can always order the car to go someplace new. And I don't believe the dream that "oh, automatic cars will just have 100% accurate street data for every road in the civilized world". First of all because we don't have this data, and even if we did, I can't imagine a car hauling several terabytes of data with it(and how would you even update it, given that our internet connections seem to be getting worse, not better every year, with new data and speed caps being introduced?). It has to be able to decide autonomously, and since it has to be programmed to do something, it's a really hard problem. A human can just "use best judgement", but a computer can't - it either follows the rules or it doesn't. And roads are not a computer simulation - they are dynamic, they change, they require the sort of input that computers are horrible at, visual recognition is dreadful, our best, cream of the top software can't tell a difference between a zebra and a sofa in a zebra print, and yet most people just wave their hand at the problems facing automatic cars saying "oh, we will just use a machine learning algorithm to watch how humans drive and then follow that". Sure. But it doesn't magically solve the problems of responsibility where a computer does something it shouldn't have.
I don't see any issue whatsoever with a car carrying around terabytes of data. That's old-think. It will never again be difficult to put startling amounts of data into devices.
And the only data that needs to be kept updated, is the local environment. The car is in one place at a time. So bandwidth should not be the issue it would seem.
You are talking about a vehicle that would have to carry dozens if not hundreds of terabytes of data. The boot would literally have to be filled with hard drives. If you are in the US you should be able to tell it to drive from one coast to the other, or if you are in Europe the car should be able to drive from Warsaw, Poland, to Aberdeen, Scotland. And what do you mean only local data needs to be updated? Why? Updating a "dumb" TomTom map takes hundreds of megabytes a month just for one country, and we are talking about much more complex data, not just a 2D map.
It needs one large read-only copy of data, plus one hard drive with updates. Those updates are limited to say a 1000 mile range (most people don't travel 100 miles from home in a month; in any case most people travel the same routes)
Sure, but if we want automatic cars to eventually replace manual cars we need to make sure they cover 100% use cases, not just most of them. Some people drive 10 miles to work and back, some people do 1000 miles a week driving through multiple countries. You need to deal with both, because nothing is stopping me from hopping in my car right this second and driving across the continent if I wish to do so. I wouldn't want an automatic car to hinder that freedom either.
The car could have some sort of wireless (4G, WiFi, etc...) to download updated/detailed data for new routes when it needs it, and store locally what it will likely need again in the future. Since the car knows where it's going, as long as it has some idea of when it will be in range of a wireless access point, it can plan when and how it will download the rest of the trip information on the way.
As long as the car has basic route information and knowledge about where it will be able to connect to servers with updated information. Cars could drive anywhere with basic road info, and download updated and more detailed information (intersection metadata, possible closures, etc...) only as needed. It doesn't always need to be in wireless range, it just needs to know when it will be so it can plan accordingly.
Obviously any driverless car is not going to rely on street markings and signage alone. It's going to have sensors in all directions and see that a vehicle is headed toward the intersection at a speed which suggests it will not be able to stop in time to avoid a collision. That doesn't require some magical machine learning algorithm, and it will probably watch for this better than humans do. We tend to assume everyone else is following the rules and often don't notice when someone is about to run the stop sign and hit us.
Uncontrolled intersections are a thing. Driverless cars will have to handle them as a matter of course, simply to be competent on the roads in the first place. They won't be able to simply assume that "no stop sign" means "no stop." This scenario is just an intersection which became uncontrolled through vandalism, rather than being built that way.
Yes! This. Some commenters have never heard of driving defensively it seems. Do they really just barrel through uncontrolled intersections if they don't see a stop sign? For that matter, do they really just unthinkingly barrel through greens, always? If so, it probably means you live outside of the city, where you can grow accustomed to trusting that nothing unexpected will happen.
In the absence of any signage, there is always an implicit "Yield." -- ie, no requirement to stop, but you should look for conflicting traffic, and preemptively slow down enough so you could stop if there turns out to be conflicting traffic.
Heck, in some places even a green light is worth defensively treating as a Yield. In Manhattan, when driving, I barrel through green lights when crossing streets while going up avenues (in center lanes), but treat it more like a Yield, slowing to a quickly-stoppable speed, when going crosstown on streets across an avenue intersection. So what if you have a green light? In practice all three road user types, pedestrians, cyclists, and motorists run reds all the time. You're supposed to drive defensively, and a good driverless car should too.
Anyway, I think driverless cars have an implicit advantage here in that their reaction times will be substantially lower. Hopefully they can use statistical evidence to figure out rules like the above ("Treat greens on 3+ lane roads as "No Stopping" sign, but on smaller roads, treat them as yields")
I don't think you seem to understand how self driving cars work, or even how computers work.
Firstly, the car itself need not have traversed the same road for it to have full information. It can be gathered from collective data from all the self driving cars. That means even if one car went through that intersection, it has some relevant data on it.
Secondly, you're handwaving that humans are better with the "use best judgement", which is pretty much unquantifiable. Not to mention humans have "used best judgement" and still managed to crash into things. How are you sure that the algorithmic path taken by the onboard computer isn't far superior to best judgement? We already know that computers outclass us in terms of reaction times, why not this?
Thirdly, you don't need to store terabytes of data to do anything in the same way you don't need to download the entire database of HN to browse the front page. Localized data would be sufficient. Even if everything is offline, the car would most likely just revert to its learning mode .. which is how most of the Google cars that are on the streets currently started off.
Lastly, you're right we aren't 100% there when it comes to distinguishing zebras and couches. But for self driving cars, we don't need that. What we need is object detection, its velocity relative to us, and data on where its heading. All of which we have become pretty good at.
Regarding your last point - we sort of do need accurate detection. What if there is a balloon on the road? I wouldn't swerve to avoid hitting a balloon, but an automatic car is going to see a round, rather large object. It might as well be a 100kg rock, hitting which would be disastrous. What is the car going to do? Stop and tell you to investigate? How about piles of leaves, cardboard boxes, large soda bottles. If you live anywhere with wind you are going to get those on the road.
As for the "best judgement" part - if a human is about to get into an accident, they are going to do what feels best. It's not necessarily the best option, but no court would ever tell you "our simulation tells us that if you reacted 0.1s earlier you could have turned the wheels 30 degrees and hit a car with two occupants instead of four, who had better survival chances". A computer is going to make an algorithmic choice, and that's what the article was about - who makes the choice as to how the car should react? The owner? The manufacturer? The police? The insurer? Should it prioritize its owner life at the potential cost of other lives? Or minimise the damage to protect the interests of its insurers?
We don't think about cars in those categories because we don't(and I would say we can't) make those choices when we drive. But computers will have to be programmed in some way. A lack of stop sign is a specific type of intersection in itself(where anyone going from the right has priority but anyone coming from the left has to yield) - so the car would need to observe and learn. I'm not saying it's impossible, but it's very hard to do.
If something can enter the road and you can't stop in time then you have already failed. Driver-less cars should never drive at a speed where they can't stop in time.
In terms of edge cases. Lidar gives a 3d image so a floating balloon is going to show up like a bird and not something to be dodged. Same is true of balls and other small fast moving objects, they are not going to ping as kids. An object on the road is going be detected in plenty of time to slow down and stop or just avoid it without hitting any thing. Or, more likely just driven over the wheels are the only contact point and clearance is high enough for many objects in the roadway.
Driver-less cars have vastly shorter stopping distances at all speeds so they can be safe at speeds humans can't. They also rarely need to hard break because they can start slowing down sooner. Dodging might be considered if traction is good and there is open space to move to.
PS: If human drivers where not killing 20,000 Americans a year then we might debate perfection. Until then, self driving cars that kill 1,000 people per month would be a massive improvement end of story.
PPS: Remember google's self driving car from last month is worse than whatever they end up putting into production. The first production cars are also going to be worse than whatever most people end up using.
"PS: If human drivers where not killing 20,000 Americans a year then we might debate perfection. Until then, self driving cars that kill 1,000 people per month would be a massive improvement end of story."
This is the biggest faulty argument in this entire discussion.
Imagine a radiotherapy machine. Statistically, due to operator errors, it delivers lethal dose of radiation to 1 patient every 10k patients. Then someone brings out a radiotherapy machine that doesn't need an operator, but statistically delivers a lethal dose of radiation every 100k patients. That machine would be banned faster than you can say litigation, even though statistically it saves lives vs. manually operated machines.
As soon as an automatic car kills the first person(be it due to faulty programming or mechanical breakdown) the manufacturer will recall all of them and the backlash will be huge, even if an average, they are still thousand times safer than manual cars.
You might think that, but if you look at say the early history of cars society decided it was perfectly OK for them to kill pedestrians. As a nation we really don't care about small numbers of people being killed if we can get something from it.
Early on flying was incredibly dangerous and that hardly stopped anything. Hell, the general public gets killed at an air show and it's a non story.
Introduction of autopilot has reduced plane accidents dramatically, yet every time one fails it's a huge thing, thousands of planes get recalled and people get fired over bad code.
People disproportionally care about things we think we should control. Like that fire of a Tesla that happened some time ago - people didn't care that all in all, statistically, electric cars are a LOT less likely to catch fire than regular ones, but media wouldn't shut up about it for weeks, highlighting how dangerous it is to drive around with a huge battery pack(even though it's nonsense).
If an automatic car has an accident, it will be a huge deal as well - because people will expect computers to be flawless, and if they make a mistake they will want the head of whoever made the computer.
Bad press != major issue. Sure the FAA grounds aircraft for a little while after defects are located untill they are fixed. They don't scrap the aircraft or ban autopilots. They also do the same thing for mechanical issues.
In tesla's case it got press as tech news, and they kept selling cars just fine. Safty issues can impact sales, but that's rational and it takes a lot more than a one off.
IMO, none of these actually support your viewpoint.
1. "Driver-less cars have vastly shorter stopping distances at all speeds"
I can understand your expectation of reduced stopping distances due to the near-0 reaction time made possible by sufficient computing power. However, I wouldn't describe the computer's advantage as anything close to a "vast" one. There is no mechanical advantage to braking that is inherent for a driverless car and the difference in stopping distance due to reaction time is less than 1 car length (~10ft) at 30mph assuming the human driver has an average reaction times (0.25s). The advantage is still limited to roughly ~25ft at 60mph. That advantage is simultaneously significant and insignificant in that incredibly common outside factors will more than account for that disparity if not controlled for in the AI's programming. Tire pressure, tread depth, brake pad thickness, rotor wear, alignment, etc are all capable of individually accounting for a ~25ft reduction in stopping distance which leads me to my second point.
2. "Driver-less cars should never drive at a speed where they can't stop in time."
Real world driving conditions present so many uncontrolled variables that a speed considered "safe" by that logic will be pathetically slow. The current state of the art in driverless cars is Google's purely autonomous design which is limited to the pace of a golf cart (25mph). The number of variables introduced by vehicle condition and maintenance alone are so great that any vehicle capable of accurately accounting for each one would be impossibly expensive to buy and crushingly expensive to maintain. While all of our cars are capable of measuring speed (the legal maximum is +/- 5mph @ 50mph in the US, but +/-1-2% is most common) and most cars can measure tire pressure (+/-15% for the most common systems which measure indirectly with the ABS sensor), they don't measure either of those things accurately according to manufacturer specs and things only get worse over time. Personally, I can't see driverless cars ever growing past that hurdle, winding up limited to inner-city commuting/taxi duty.
3. "Same is true of balls and other small fast moving objects, they are not going to ping as kids."
This is actually one of the best examples of why speeds will have to remain slow enough to guarantee a negligible stopping distance.
If a human sees a ball roll into the street from behind an obstruction, they will reasonably deduce that there is a likelihood that a kid is chasing after it and slow down before the danger presents itself. An AI will see the balloon or ball and it will have to treat it as either a complete non-issue or as a mortal danger. If it treats it as a non-issue, it will have to be in a constant state of readiness to stop in the smallest reasonable distance, that means limiting speeds to a relatively glacial pace. If it treats it as a mortal danger, it will constantly produce false positives which leads to frustrated passengers when it brakes hard for every trashbag that blows across a 70mph highway.
That's one of my main criticisms of driverless cars in their current state. They are measurably superior in the areas in which human drivers are most flawed, yet their logic leads them to make decisions that no human would ever consider as logical. Even the most state of the art AI isn't even close to being able to account for the nearly infinite number of exceptions that stem from the chaos of the real world and which require an action contradictory to a principle of placing safety first. Just look at the first "unforced error" reported by the Google program where the AI's logic misjudged the actions of a bus due to logic which defined exceptions around the distance needed to merge on to a road. It just went for the merge, turning straight into the side (pretty much in the middle) of a bus moving at 15mph. No human would have made that decision.
1. Don't confuse optimal human reaction times with real world conditions. Drivers don't have unwavering focus on the path ahead. Accidents are the exception when things go wrong, so the median accident reaction time would be far higher than under optimal conditions and really hard to measure. If a self driving car can't safely stop then it really should complain and just slow down. I can see "Speed reduced Tires balding" really would get most people into the shop and my car already tracks tire pressure and breaking. Add GPS for absolute location and it will notice if there is a problem. Listing for weather reports is also a no brainier.
2. Google is being conservative with several cars being tested at much higher speeds, don't forget in the computer world being 1/3 as fast is really far less of a jump than you might think. In terms of accuracy GPS can auto correct distance traveled over time to get much higher accuracy. Much like how most computers clocks are far closer to the correct time than their cheep HW clock would suggest.
3. Your thinking like a human. Kids are really really slow at 10mph (which is fast for a kid) 7.5 feet is over 1/2 a second. So, your car can drive like there is always someone about to jump out from every intersection and sprint across the road without being all that slow. Depending on sensor location they can also see under other cars. ED: the Front drivers side headlight is a much better location to see around parked cars than where a driver sits.
Worst case self driving cars might drive slow when stuff is right next to the road. But, if I can be productive and safe it's a non issue as most driving is not in those conditions and I should slow down anyway.
" If something can enter the road and you can't stop in time then you have already failed. Driver-less cars should never drive at a speed where they can't stop in time."
So, what happens if something blows into or falls on a motorway? A car hurtling down the road at 70mph is going to have a hell of a time stopping for an unexpected obstacle, human controlled or otherwise.
Not sure what your point is? Cars are not going to be dodging bullets or anything.
But, on a highway where it's safe to do 70MPH a deer running across the road is plenty slow enough to be stopped for. In terms of birds and bunny's just hitting them is the normal human response.
IMO, getting out into the once per year in the US edge cases is kind of missing the point. Ok, if your driving in a hurricane and a roof falls onto your car then yea your going to have a bad day meh.
>A computer is going to make an algorithmic choice, and that's what the article was about - who makes the choice as to how the car should react?
Google has already said that this kind of decision making will not be a part of their cars' system. The car's reaction to situations is based on the rules that we all agreed on when we passed laws surrounding how humans are to behave on the road.
If the car encounter as obstacle, it will brake. It will not calculate death probabilities. It does not need to.
This is not the issue with that statement at all, and for the record I'm skeptical of cars sharing data to augment what they see (because this leaves them open to attacks that are unique to self-driving cars).
The problem with the statement is that it, as the GP post says, ignores the fact that a human driving a car is no less likely to barrel through an intersection with a missing stop sign.
There are a few other low level problems with this scenario as well, mostly hinging on the weird idea that self-driving cars are just going to barrel into an intersection without a stop sign as if there are no other cars going through it. This is why they have radar/lidar/whatever else.
There's a lot of clicks & eyeballs associated with FUD articles about next-gen cars at the moment. (autonomous, electric, or both)
If you want to draw in a techie audience with clickbait, there's not much better you can do at the moment than writing about how Tesla has some sort of unforeseen flaw whereby both their electric & autonomous tech is doomed to failure.
It seems to be really hard to get people to compare driverless cars with real humans, and not imaginary perfect drivers.
For another example, there's the nigh ubiquitous question of how to handle ethical dilemmas, like a crash where you have a choice between crushing a nun and crushing a baby carriage. Nobody ever seems to consider the fact that human drivers will almost never consider the ethics of the situation when choosing an outcome, and in fact probably won't deliberately choose any outcome at all, but will merely brake and steer by reflex.
If you're ever driving in a situation where it is possible for a nun or baby carriage to jump in front of your car without any warning, you should be driving at 30km/h (20mph) or less. This gives you a stopping distance of approximately 1 car length, so stopping is always a better option than trying to steer around an obstacle. It also means that any collision with a pedestrian is highly unlikely to be fatal.
Of course, having automated cars driving at 30km/h through the suburbs is going to piss everybody off, but that's better than dead babies.
Grew up on a road where your choice is more likely to be one kid vs. another kid. Nobody drove 20 mph on that road. At least driverless cars will respect the speed limit. Human drivers do no such thing. While texting.
I'd wager a driverless car can afford to go a little faster, given its reaction speed is likelier much, much faster than anything a human can put out.
In fact, a driverless car may be capable of estimating its own stopping distance given road conditions and other measurements, account for some margin of error, and then use that to calculate how fast it can go down a road, while still adhering to the posted speed limits.
If the human in the car isn't impatiently driving it, they'll probably be happier with a slower, safer, and more comfortable ride while they fiddle with their phone or do work on their laptop. Speed has no business, frankly, on residential drives. I don't think it even helps you get anywhere much faster -- it's more about venting schedule anxiety onto your gas pedal.
It's not just about reaction time; 30 km/h is the limit where you can collide with a pedestrian and be fairly assured that the pedestrian will live. Above that speed survival rates drop dramatically.
The point of that often quoted ethical dilemma is that a human driver can't choose at all. With driverless cars at one point someone has to make a deliberate decision on how to act in these situations. Not in a split-second but rather in a long process. This results in different ethical considerations.
Sure, a human driver doesn't take any of the relevant factors into account. But a driverless car does and that's the difference.
That's my point. People's expectations for driverless cars are vastly higher for no particularly good reason. Sure, a driverless car could potentially do a lot better than a human, but if it merely does as well as a human then it's already good enough here. You can slice the gordian knot of the ethical dilemma by saying "when presented with two bad choices, we'll pick whichever one is closer to the car's path" and it's fine. But people talk about it as a difficult problem to overcome because they're unwilling to accept human-level performance.
"It seems to be really hard to get people to compare driverless cars with real humans, and not imaginary perfect drivers."
I think this responsibility falls squarely on the manufacturers of autonomous vehicles.
I think only the people creating the technology really understand the strengths and weaknesses of software drivers compared to human drivers, and are responsible for educating law makers and the general public if they want their products to be accommodated in our laws, road ways, and market places.
I think journalists are responsible for understanding the facts and implications of what they discuss, regardless of how good the various players are at explaining it.
Yes, the manufacturers have a much stronger foundation from which to figure this stuff out, but that doesn't mean the rest of us can't stop doing silly things like acting as if human drivers were perfect when talking about potential autonomous vehicle systems.
I think it's very difficult, though, to know what state of the art software driven vehicles are good at and not so good at, though. Maybe the missing-stop-sign-intersection was a problem in version 0.7.1 but fixed in patch 0.7.1.4.
Any good articles by engineers describing where current autonomous cars are still worse than humans? And which issues are just "bug that hasn't been fixed yet" vs. "we're still not sure how to solve this one in principle".
Agree, though, that a journalist should be interviewing real engineers and trying to summarize, paraphrase, and edit what he hears to the best of his or her ability, instead of just speculating.
Humans would also improve with removed signage. There's quite a bit of research now that accident rates go down the more you remove street furniture and lines. Same with traffic lights - see the caution people use when the lights are out of order.
I'd hope that a self-driving car has the meta data on board rather than just relying on seeing the sign. I'd also hope that they're designed to treat all junctions with caution, not just those with missing signage.
Indeed, there is an intersection near where I live where the stop sign is removed and then later replaced (seemingly randomly, but probably not) every few weeks.
When there is a stop sign, I stop. When there is no stop sign, I don't stop.
Those are the rules of the road. Stopping for a stop sign that doesn't exist is breaking those rules, and thus putting other people in danger.
The fact is that GPS navigation is a mature field (I.e. unlikely to get much better very soon), and is still very failure-prone. I just took an Uber this morning and twice the GPS got confused between the street I was going to and the on-ramp to the highway (the two roads were narrow and unusually close together). Twice, the driver followed the GPS instead of looking with his eyes at the road (the on ramp clearly looked like an on ramp). I run into these sorts of things 2-3 times a month.
An uncontrolled intersection means yield to traffic approaching from your passenger side for intersecting roads of equal importance, or vehicles on the minor road yield to through traffic on the major road. Yield signs usually just formalize this implicit intersection control. Taking out one stop sign is usually not an imminent hazard to traffic, but taking out two from perpendicular approaches might cause a problem. Removing all of them would actually be slightly more safe than a bad configuration of signs.
I expect my driverless car to be able to disobey signs when I indicate that it would be appropriate or necessary to do so.
I may one day enforce this by carrying around two poles that bear "stop", "do not enter + wrong way", "no u-turn", "do not reverse + severe tire damage", and "no stopping any time" signs on them, and creating pop-up paradox traps for unwary driving AIs.
Cities will become completely Undriveable and possibly off limits to cars once the majority are self driving.
Here's why; pedestrians will know that all cars will stop for them and people will just start crossing anytime they want and even start simply walking in the street. There might be some jaywalking enforcement attempts but ultimate they'll fail.
Frankly if you're hitting people just because they aren't in a crosswalk while you're driving, you have no business driving now, let alone in the future. As the controller of the 2.5 ton metal machine, you are responsible for making sure you don't hit things or people.
With the saturation of safety gear on newer cars you really have no excuse anyhow, the newer ones even slam on the brakes FOR you when they detect an obstacle.
People already feel entitled to block traffic they don't think will kill them.
One time someone was standing in the bike lane attempting to cross a 3 lane road against the light. She seemed upset that I was going to ride my bike through the green light anyway, and told me that you have to always stop for pedestrians in the crosswalk regardless of the traffic signal. Unfortunately, pedestrians aren't allowed to enter the crosswalk except when they have a walk signal (that flashing red hand means you're not allowed to cross, though I've never seen a single person follow that rule).
As a cyclist I think we should ban both walking and driving ;)
> and told me that you have to always stop for pedestrians in the crosswalk regardless of the traffic signal
Well she's right; of course you have to stop for pedestrians in the crosswalk. You can't wilfully hit someone and just say 'my right of way' and be all in the clear.
I think in the UK pedestrians actually always have priority, as long as they are travelling in a straight line and you as a driver are making some kind of turn. So for example a road joining another road at a T junction, pedestrians crossing the joining road have priority over cars, regardless of lights or markings.
I think your response is missing the point. Of course you have to avoid hitting someone even if they're technically in the wrong. The point was that people feel entitled to do things without even a basic understanding of the law (or even what's sensible).
Indeed. The fact that cars will swerve to avoid you doesn't mean that you should delay traffic because you feel like crossing now rather than in 30 seconds. The crossing of streets are coordinated by an agreed-upon mechanism. Try to observe it, so everyone gets their fair chance to make forward progress.
> You can't wilfully hit someone and just say 'my right of way' and be all in the clear.
Willfully perhaps, but if it's accidental and all other circumstances being equal I would say that's the pedestrian's fault. I don't understand this logic that because the pedestrian is ignorant about how traffic law works that they're exempt from following it, yes the driver should be paying attention at all times and be ready for anything to happen, but if someone breaks a don't walk light and just wanders into traffic, they should be held liable for the resulting accident.
Why would enforcement attempts fail? They don't work right now because nobody really cares, jaywalking is almost never a real problem. If it starts seriously affecting transportation, laws will change and enforcement will increase.
Some places are like this already but people manage, in Brighton in the UK people walk in the road all the time... in London, if you try this, well - don't try it.
As someone who works in London, I can definitely say that a lot of people will walk straight into the road at the first slightly clear opportunity. In the the City of London part, your car being stopped for about a minute seems to be enough for the large group of waiting pedestrians to try and walk around it.
I think we know (i.e. 90% of the time) that people will stop for us if we walk out in front of them with ample notice and conditions... It's more from a norms / politeness perspective that we don't; it's simply not accepted that you can just walk into the road when you want to get across because (most of us) we don't want to be judged by others for behaving recklessly, either by the driver we're forcing to stop or the other pedestrians.
There will most often than not still be people in these driverless cars that we will irritate, and pedestrians that we fear will scorn us. However, I think you make a good point about non-passenger vehicles. Perhaps we will be more likely to walk out in front of an unoccupied Uber that we know will stop for us on an empty street where no one will judge us — but this won't affect the busy streets / scenarios where it's most important that I mention above.
I go out of my way not to step in front of a car if it is on a trajectory that would hit me. It's a "never trust the driver" mantra.
These aren't busy city streets where I'm making that choice, but I'm not getting to the point where I'm being considerate of the driver, I simply don't want to create the situation where I am dependent on their noticing me.
I do the same, but will walk as if I'm about to (as long as I'm in a crosswalk, where drivers are supposed to yield to pedestrians). I can stop walking way faster than drivers can stop their cars, so they basically have to lose the game of "chicken" or be willing to run over pedestrians in crosswalks. Only once they've visually committed themselves to stopping do I take that last step into the path of the vehicle.
It's a little douchy, but goddamnit, California law says I've got right-of-way in crosswalks. When you're holding a mutex, what you do is get rid of it ASAP and get out of the way.
"California law says I've got right-of-way in crosswalks"
That's for the safety of the pedestrian in order to avoid confusion and make liability strictly fall on the shoulders of the party with the 2 killing machine.
Regardless of my legal right-of-way, though, I try not to be a jerk about it.
If you're about to cross a crosswalk, you're contributing to the attentional noise that drivers have to deal with. IMHO, just get out of the way so that fewer people have to make sure that they aren't going to hit you.
I generally agree that doing a dance entering the crosswalk is unhelpful, but there are lots of cases where you can stand back and wait to enter the crosswalk in order to avoid making a car slam on its breaks.
Nothing douchey about it. Car drivers are taking an enormous amount of public resources and endangering the public while they do it. They can wait a few seconds so that you don't have to wait indefinitely for them in crosswalks.
As a tangent, I think people are severely underestimating how high the percentage of non-passenger vehicles will go up when we're driverless. I'm not talking about empty passenger vehicles looking for passengers, rather goods being shuttled to and fro. Look at all the startups for bringing you food and various other goods that are semi-profitable even with a human wage. History and common sense show that making something cheaper will mean we make more use of it. The driverless future will be filled with autonomous vehicles moving goods around our cities. In the extreme we may even lose apartment kitchens in dense areas because shuttling your meals to you may be cheaper.
It's more from a norms / politeness perspective that we don't; it's simply not accepted that you can just walk into the road when you want to get across because (most of us) we don't want to be judged by others for behaving recklessly
Anyone who has ever lived (or otherwise spent a lot of time driving) in a bad neighborhood knows intimately that this is true. People walk in the street in rough places. And there aren't any norms compelling them to get out of the way.
Nah I think there's always a little bit of fear about that 10% of the time. Humans are unpredictable- what if they don't see you until it's too late. What if they want to give you a "scare" and screech to a halt right at your feet- and what if they get the distance wrong and wind up hitting you. Robot cars on the other hand, while they might fail on a rare occasion, will have very predictable behavior.
When you walk up to a doorman, you make eye contact and look for him to open the door. When you walk up to an automatic door, you have been trained to know it will open for you.
Run an experiment with a bad doorman who doesn't open doors, and an automatic door that doesn't open, and you'll have a lot more people running into the automatic door.
This doesn't happen in countries unlike the US where "jaywalking" is not a crime or even a concept. I think even people that know intellectually that the cars should stop aren't going to push their luck in traffic traveling at speed.
Jaywalking is definitely a crime and a concept in the U.S. In urban environments where speed limits are usually lower, the laws are generally unenforced (source: I'm a serial jaywalker) however you definitely can get a ticket crossing a busy highway.
I originally interpreted his phrasing the same way as you, but I think it's a bracketing thing:
> This doesn't happen in countries unlike the US where "jaywalking" is not a crime or even a concept.
I think it should be "This doesn't happen in (countries unlike the US) where "jaywalking" is not a crime of even a concept" rather than how I initially read it "This doesn't happen in countries unlike (the US where "jaywalking is not a crime or even a concept.)"
More likely that cars will automatically report jaywalkers and that police, due to a lack to traffic tickets for cars, will start giving tickets to those on foot.
That's an interesting point. I think you may see something similar to tourist districts today where the streets are closed to traffic outside of certain hours. People drive near the area and continue on foot.
Autonomous vehicles may be even better for this, as a person could be dropped off at the closest accessible point, then the vehicle retreats to a lower-density area.
There are already places where people walk like this, and they still have cars (even if they're a bit less effective).
With that said, I hope they phase out human driving (or require similarly ex(p|t)ensive insurance policies) once SDCs become robust; the joy of driving isn't enough, IMHO to build the entire infrastructure and compromise safety.
The one thing the author doesn't seem to be taking into account is that the cities are much easier for them to change as needed. The infrastructure is more dense and (usually) more orderly designed, and access to electricity to add beacons, as well as access to all kinds of networking resources already in place for traffic work is pre-existing. Augmenting cities for self driving cars is going to be the bunny hill compared to augmenting the interstate highway system to them.
I suspect you haven't been in a big city much if you think thousands of stop-and-start intersections would be easier to automate than a far fewer number of continuous lanes and offramps that work via simple merging.
Automated interstate driving is nearly a solved problem already. My car will easily go for hours hands-off on most interstates. The only infrastructure you need is clearly marked lines, which already exist on most of them. There's no way that cities are easier to change as needed, when interstates don't need to change in the first place.
Did the author actually ask the makers of self-driving cars whether these are real problems? Many of the problems listed in the article seemed relatively easy to solve (compared to the difficulty of building a self-driving car).
I actually had a long discussion about this with an Uber driver the other day. Given that this article predates that discussion, I would imagine he was sourcing a lot of his arguments from this article.
My position in this discussion is that yes, infrastructure improvements are necessary, but they'll be relatively straight forward in the US, as basically every city in the US was designed to support automobile traffic.
Self driving cars are going to have a much tougher row to hoe in Europe and Asia, where city infrastructure predates the automobile by centuries/millenia
>My position in this discussion is that yes, infrastructure improvements are necessary, but they'll be relatively straight forward in the US, as basically every city in the US was designed to support automobile traffic.
This is not to say that it's impossible, but US cities barely manage to get the infrastructure upgrades and maintenance they need now - Google et. al. have a lot of money, but I'm not confident they can play bureaucracy well enough to get some of the infrastructure improvements necessary for either large fleets of SDC's. I admit, the idea of replacing all the individually owned cars on the road with a fleet of summonable automobiles is very appealing; it just seems like there are a lot of intermediary steps that are less about technology and money than they are about politics and personal interests.
I'm more optimistic that the NTSA will rise to meet the challenge because there is so much benefit and momentum (thank you Elon Musk) to making them work.
My favorite article on the subject[1] extrapolates how much impact autonomous vehicle technology will have. I'd actually be pretty surprised if it hadn't come across the desks of the GM executives involved in the half a billion dollar Lyft investment[2]. The Obama administration seems to be making a real effort to lay the groundwork for the necessary infrastructure and regulatory resources[3].
>And in cities where it is customary for human drivers to anticipate the red light turning to green by inching into the intersection prematurely, will the driverless automobile allow for the custom?
I can't believe they are describing this as a custom rather than a dangerous behavior born of impatience. I grew up in a place where people did this and it only took one close call with a clearly drunk driver flying through a red across the intersection in front of me to make me stop for good. Why would anyone design a self driving vehicle to emulate this behavior?
I've wondered several times, how could a human on the road interact with an autonomous car? For example, if a lane is currently closed and alternate traffic is in place on the other lane, how can a traffic cop instruct the car to stop there for a while and then proceed to cross a double white line and move in the usually-wrong direction? Or, how can you explain a car that a bridge is blocked and it has to take the next one?
Articles like this feel like they were ripped from another century.
"Why cities aren't ready for the horseless carriage."
... Sure, they may not be. But that doesn't mean that change will stop, or that they have to be ready overnight. It means that we'll have to deal with these things eventually, and I'm betting it's sooner rather than later.
I just don't think that we as a society are ready for the driverless car. We are not ready to give up the freedom and the semi-anonymous nature of driving. I know you'll that with ubiquitous cameras that we are already being tracked, but that is different from the continuous, high-fidelity tracking that will be in place on these cars. I trust neither the private companies nor the government to keep this data secure. I expect that within my driving lifetime that I'll be forced to choose between lower insurance (for driverless cars) and driving myself. That makes me sad.
And consider, if you will, how this can become the modern version of "Red Barchetta" if you did choose to keep driving yourself.
I'm absolutely with you. The weather's finally warming up, so I spent most of this past weekend doing maintenance on my motorcycle for the riding season. In a couple of decades will riding a motorcycle on public roads even be an option anymore? Will all motorcycle riding be relegated to isolated tracks, to which we'll have to ferry our motorcycles on autonomous transport?
I'd rather not ride at all at that point, and that makes me sad as well.
You can have self driving cars where you can turn the system on and off like you can on the Tesla S. That would work for me. I can't see them banning human driving in a hurry.
I have to complain that the WSJ's title is completely backwards. Petroski's article is about the sensible topic that self driving cars aren't ready for the typical real-world chaos. In fact he glossed over Audi's "99% of the trip was autonomous" claim -- that 99% was on freeways which while hard, are well structured.
I can't wait for self driving cars for a jillion reasons but I am dismayed when advocates claim that the infrastructure must change to support them. That's completely backwards: our tools should support us, not the other way around.
That is a losing game. Years ago, I worked on a DARPA project that would let radios opportunistically use available frequencies instead of fixed ones. Doing this with a totally uncooperative environment, relying only on the wits of the software, is really hard. Like still in development after more than a decade hard. Like Microsoft and Google gave up on solving an easier problem hard.
Self-driving cars in an uncooperative environment is probably an order of magnitude harder than that. Demanding that they work in an uncooperative environment results in a problem that is probably intractably hard.
>> Self-driving cars in an uncooperative environment is probably an order of magnitude harder than that. Demanding that they work in an uncooperative environment results in a problem that is probably intractably hard.
That's why I'm skeptical of self-driving cars becoming widespread in the U.S. in the next decade. Driving in an unknown environment over millions of miles of roads is going to be a huge challenge if the infrastructure doesn't change, but I can't figure out which political party is going to support the massive infrastructure project. The early adopters of this technology are probably going to be transportation companies. Democrats are going to have a hard time advocating for spending money that throws truck and taxi drivers out of work. And good luck getting Republicans to spend additional money on anything.
I hope that google and/or the car companies pull it off, but I don't think that we should hold our breath waiting for the world to adjust to self-driving cars. The hardware and software will have to do 99% of the work.
My prediction (I'm well aware this is worthless, as nobody cares about random-internet-guy's predictions enough to follow up) is that there will be a handful of cities that will try their best to orient themselves around self driving cars, and like the cities before them that tried to orient themselves around cars, they will fail. Most of the fanciful predictions from self-driving car futurism are incredibly naive and fail to take into account basic facts about how cities function from a systems perspective.
1) Traffic will disappear because cars can stop faster and travel close together. At best there are a tiny minority of driving scenarios where driving closer together or stopping quicker are the system bottlenecks. Rush hour traffic is not one of them. At best, you will drive faster on highways, but that will just get you to your exit faster, where you will spend more time waiting than you used to because intracity bottlenecks haven't changed. Traffic lights will still continue to make tradeoffs between latency and throughput, and self driving cars can't take away that tradeoff. And cars will continue to drive slow enough to stop for typical city obstructions (like cyclists, pedestrians, etc.), even if they have a few milliseconds faster of a reaction time. And you're still going to have typical rush hour volumes of people coming from multiple different places converging on a single area.
2) Haven't you seen the simulation where self driving cars don't need traffic signals anymore cause they can drive right through without stopping or hitting anybody? Guess what, it's a simulation...one that conveniently ignored almost every other relevant factor about how the world works. Like pedestrians. Dogs. Joggers. Cyclists. Handicapped people. Gridlocks that happen due to bottlenecks on perpendicular streets. Street protests. Weather conditions. In fact, the only way this idea feasibly works is if you are in a remote intersection in the middle of somewhere so unimportant that it doesn't have any other obstacles or traffic types, in an area where non-driverless cars are banned. And guess what...those situations get along quite fine with traffic lights.
3) But you won't have to circle around the block endlessly to find parking! Actually, your situation isn't much better, and if you see a rise in commuting by cars it could get much worse. Your traffic lanes will be continuously blocked by people getting in and out of their cars. Think of the airport arrivals and departures loops...nobody there has a need for parking, and yet still they are a traffic nightmare because everybody is either getting in a car or getting out of a car, and you have cars blocking traffic looking for their specific passenger, and you have cars circling around the block anyway because there is nowhere to stop legally.
I could go on and on. Traffic is the external manifestation of a system that is far more complex than streets with a few bad drivers clogging shit up. It is a manifestation of the interaction between cars and non-cars, economies and the humans that contribute to them, all with different goals and objectives, all trying to use a shared but very scarce resource. It isn't going away. Ever. The best you can do is find ways to move more people through it faster, and from that perspective the geometric reality of an automobile is grim. No matter how optimized for ommitting the driver it may be, a car will always take up exponential amounts of space compared to public transit systems.
But that isn't a reason to not pursue them. They should still be pursued on safety grounds alone. They just won't answer your prayers for a traffic-free city.
The author naively assumes that driverless cars are going to be replacements of human drivers in most ways. This is an extremely ignorant view that lacks a lot of imagination. Everyone actually working on driverless cars has been thinking of all these potential gotchas for years and have been designing and engineering with these gotchas in mind.
104 comments
[ 2.9 ms ] story [ 59.6 ms ] threadBut so would human drivers. In fact, a self-driving car may actually do better in this scenario because it would have metadata stating that the last 3,000 times a car encountered this intersection, there was a stop sign here.
And the only data that needs to be kept updated, is the local environment. The car is in one place at a time. So bandwidth should not be the issue it would seem.
As long as the car has basic route information and knowledge about where it will be able to connect to servers with updated information. Cars could drive anywhere with basic road info, and download updated and more detailed information (intersection metadata, possible closures, etc...) only as needed. It doesn't always need to be in wireless range, it just needs to know when it will be so it can plan accordingly.
In the absence of any signage, there is always an implicit "Yield." -- ie, no requirement to stop, but you should look for conflicting traffic, and preemptively slow down enough so you could stop if there turns out to be conflicting traffic.
Heck, in some places even a green light is worth defensively treating as a Yield. In Manhattan, when driving, I barrel through green lights when crossing streets while going up avenues (in center lanes), but treat it more like a Yield, slowing to a quickly-stoppable speed, when going crosstown on streets across an avenue intersection. So what if you have a green light? In practice all three road user types, pedestrians, cyclists, and motorists run reds all the time. You're supposed to drive defensively, and a good driverless car should too.
Anyway, I think driverless cars have an implicit advantage here in that their reaction times will be substantially lower. Hopefully they can use statistical evidence to figure out rules like the above ("Treat greens on 3+ lane roads as "No Stopping" sign, but on smaller roads, treat them as yields")
As for the "best judgement" part - if a human is about to get into an accident, they are going to do what feels best. It's not necessarily the best option, but no court would ever tell you "our simulation tells us that if you reacted 0.1s earlier you could have turned the wheels 30 degrees and hit a car with two occupants instead of four, who had better survival chances". A computer is going to make an algorithmic choice, and that's what the article was about - who makes the choice as to how the car should react? The owner? The manufacturer? The police? The insurer? Should it prioritize its owner life at the potential cost of other lives? Or minimise the damage to protect the interests of its insurers?
We don't think about cars in those categories because we don't(and I would say we can't) make those choices when we drive. But computers will have to be programmed in some way. A lack of stop sign is a specific type of intersection in itself(where anyone going from the right has priority but anyone coming from the left has to yield) - so the car would need to observe and learn. I'm not saying it's impossible, but it's very hard to do.
In terms of edge cases. Lidar gives a 3d image so a floating balloon is going to show up like a bird and not something to be dodged. Same is true of balls and other small fast moving objects, they are not going to ping as kids. An object on the road is going be detected in plenty of time to slow down and stop or just avoid it without hitting any thing. Or, more likely just driven over the wheels are the only contact point and clearance is high enough for many objects in the roadway.
Driver-less cars have vastly shorter stopping distances at all speeds so they can be safe at speeds humans can't. They also rarely need to hard break because they can start slowing down sooner. Dodging might be considered if traction is good and there is open space to move to.
PS: If human drivers where not killing 20,000 Americans a year then we might debate perfection. Until then, self driving cars that kill 1,000 people per month would be a massive improvement end of story.
PPS: Remember google's self driving car from last month is worse than whatever they end up putting into production. The first production cars are also going to be worse than whatever most people end up using.
This is the biggest faulty argument in this entire discussion.
Imagine a radiotherapy machine. Statistically, due to operator errors, it delivers lethal dose of radiation to 1 patient every 10k patients. Then someone brings out a radiotherapy machine that doesn't need an operator, but statistically delivers a lethal dose of radiation every 100k patients. That machine would be banned faster than you can say litigation, even though statistically it saves lives vs. manually operated machines.
As soon as an automatic car kills the first person(be it due to faulty programming or mechanical breakdown) the manufacturer will recall all of them and the backlash will be huge, even if an average, they are still thousand times safer than manual cars.
Early on flying was incredibly dangerous and that hardly stopped anything. Hell, the general public gets killed at an air show and it's a non story.
People disproportionally care about things we think we should control. Like that fire of a Tesla that happened some time ago - people didn't care that all in all, statistically, electric cars are a LOT less likely to catch fire than regular ones, but media wouldn't shut up about it for weeks, highlighting how dangerous it is to drive around with a huge battery pack(even though it's nonsense). If an automatic car has an accident, it will be a huge deal as well - because people will expect computers to be flawless, and if they make a mistake they will want the head of whoever made the computer.
In tesla's case it got press as tech news, and they kept selling cars just fine. Safty issues can impact sales, but that's rational and it takes a lot more than a one off.
IMO, none of these actually support your viewpoint.
1. "Driver-less cars have vastly shorter stopping distances at all speeds"
I can understand your expectation of reduced stopping distances due to the near-0 reaction time made possible by sufficient computing power. However, I wouldn't describe the computer's advantage as anything close to a "vast" one. There is no mechanical advantage to braking that is inherent for a driverless car and the difference in stopping distance due to reaction time is less than 1 car length (~10ft) at 30mph assuming the human driver has an average reaction times (0.25s). The advantage is still limited to roughly ~25ft at 60mph. That advantage is simultaneously significant and insignificant in that incredibly common outside factors will more than account for that disparity if not controlled for in the AI's programming. Tire pressure, tread depth, brake pad thickness, rotor wear, alignment, etc are all capable of individually accounting for a ~25ft reduction in stopping distance which leads me to my second point.
2. "Driver-less cars should never drive at a speed where they can't stop in time."
Real world driving conditions present so many uncontrolled variables that a speed considered "safe" by that logic will be pathetically slow. The current state of the art in driverless cars is Google's purely autonomous design which is limited to the pace of a golf cart (25mph). The number of variables introduced by vehicle condition and maintenance alone are so great that any vehicle capable of accurately accounting for each one would be impossibly expensive to buy and crushingly expensive to maintain. While all of our cars are capable of measuring speed (the legal maximum is +/- 5mph @ 50mph in the US, but +/-1-2% is most common) and most cars can measure tire pressure (+/-15% for the most common systems which measure indirectly with the ABS sensor), they don't measure either of those things accurately according to manufacturer specs and things only get worse over time. Personally, I can't see driverless cars ever growing past that hurdle, winding up limited to inner-city commuting/taxi duty.
3. "Same is true of balls and other small fast moving objects, they are not going to ping as kids."
This is actually one of the best examples of why speeds will have to remain slow enough to guarantee a negligible stopping distance.
If a human sees a ball roll into the street from behind an obstruction, they will reasonably deduce that there is a likelihood that a kid is chasing after it and slow down before the danger presents itself. An AI will see the balloon or ball and it will have to treat it as either a complete non-issue or as a mortal danger. If it treats it as a non-issue, it will have to be in a constant state of readiness to stop in the smallest reasonable distance, that means limiting speeds to a relatively glacial pace. If it treats it as a mortal danger, it will constantly produce false positives which leads to frustrated passengers when it brakes hard for every trashbag that blows across a 70mph highway.
That's one of my main criticisms of driverless cars in their current state. They are measurably superior in the areas in which human drivers are most flawed, yet their logic leads them to make decisions that no human would ever consider as logical. Even the most state of the art AI isn't even close to being able to account for the nearly infinite number of exceptions that stem from the chaos of the real world and which require an action contradictory to a principle of placing safety first. Just look at the first "unforced error" reported by the Google program where the AI's logic misjudged the actions of a bus due to logic which defined exceptions around the distance needed to merge on to a road. It just went for the merge, turning straight into the side (pretty much in the middle) of a bus moving at 15mph. No human would have made that decision.
2. Google is being conservative with several cars being tested at much higher speeds, don't forget in the computer world being 1/3 as fast is really far less of a jump than you might think. In terms of accuracy GPS can auto correct distance traveled over time to get much higher accuracy. Much like how most computers clocks are far closer to the correct time than their cheep HW clock would suggest.
3. Your thinking like a human. Kids are really really slow at 10mph (which is fast for a kid) 7.5 feet is over 1/2 a second. So, your car can drive like there is always someone about to jump out from every intersection and sprint across the road without being all that slow. Depending on sensor location they can also see under other cars. ED: the Front drivers side headlight is a much better location to see around parked cars than where a driver sits.
Worst case self driving cars might drive slow when stuff is right next to the road. But, if I can be productive and safe it's a non issue as most driving is not in those conditions and I should slow down anyway.
So, what happens if something blows into or falls on a motorway? A car hurtling down the road at 70mph is going to have a hell of a time stopping for an unexpected obstacle, human controlled or otherwise.
But, on a highway where it's safe to do 70MPH a deer running across the road is plenty slow enough to be stopped for. In terms of birds and bunny's just hitting them is the normal human response.
IMO, getting out into the once per year in the US edge cases is kind of missing the point. Ok, if your driving in a hurricane and a roof falls onto your car then yea your going to have a bad day meh.
Google has already said that this kind of decision making will not be a part of their cars' system. The car's reaction to situations is based on the rules that we all agreed on when we passed laws surrounding how humans are to behave on the road.
If the car encounter as obstacle, it will brake. It will not calculate death probabilities. It does not need to.
The problem with the statement is that it, as the GP post says, ignores the fact that a human driving a car is no less likely to barrel through an intersection with a missing stop sign.
There are a few other low level problems with this scenario as well, mostly hinging on the weird idea that self-driving cars are just going to barrel into an intersection without a stop sign as if there are no other cars going through it. This is why they have radar/lidar/whatever else.
If you want to draw in a techie audience with clickbait, there's not much better you can do at the moment than writing about how Tesla has some sort of unforeseen flaw whereby both their electric & autonomous tech is doomed to failure.
For another example, there's the nigh ubiquitous question of how to handle ethical dilemmas, like a crash where you have a choice between crushing a nun and crushing a baby carriage. Nobody ever seems to consider the fact that human drivers will almost never consider the ethics of the situation when choosing an outcome, and in fact probably won't deliberately choose any outcome at all, but will merely brake and steer by reflex.
Of course, having automated cars driving at 30km/h through the suburbs is going to piss everybody off, but that's better than dead babies.
In fact, a driverless car may be capable of estimating its own stopping distance given road conditions and other measurements, account for some margin of error, and then use that to calculate how fast it can go down a road, while still adhering to the posted speed limits.
"He won't get there any faster"
(Yes, he will. It's pretty simple to work out exactly how much faster actually)
"He won't get there any faster"
(Yes, he will. It's pretty simple to work out exactly how much faster actually)
Sure, a human driver doesn't take any of the relevant factors into account. But a driverless car does and that's the difference.
I think this responsibility falls squarely on the manufacturers of autonomous vehicles.
I think only the people creating the technology really understand the strengths and weaknesses of software drivers compared to human drivers, and are responsible for educating law makers and the general public if they want their products to be accommodated in our laws, road ways, and market places.
Yes, the manufacturers have a much stronger foundation from which to figure this stuff out, but that doesn't mean the rest of us can't stop doing silly things like acting as if human drivers were perfect when talking about potential autonomous vehicle systems.
Any good articles by engineers describing where current autonomous cars are still worse than humans? And which issues are just "bug that hasn't been fixed yet" vs. "we're still not sure how to solve this one in principle".
Agree, though, that a journalist should be interviewing real engineers and trying to summarize, paraphrase, and edit what he hears to the best of his or her ability, instead of just speculating.
I'd hope that a self-driving car has the meta data on board rather than just relying on seeing the sign. I'd also hope that they're designed to treat all junctions with caution, not just those with missing signage.
When there is a stop sign, I stop. When there is no stop sign, I don't stop.
Those are the rules of the road. Stopping for a stop sign that doesn't exist is breaking those rules, and thus putting other people in danger.
The fact is that GPS navigation is a mature field (I.e. unlikely to get much better very soon), and is still very failure-prone. I just took an Uber this morning and twice the GPS got confused between the street I was going to and the on-ramp to the highway (the two roads were narrow and unusually close together). Twice, the driver followed the GPS instead of looking with his eyes at the road (the on ramp clearly looked like an on ramp). I run into these sorts of things 2-3 times a month.
I expect my driverless car to be able to disobey signs when I indicate that it would be appropriate or necessary to do so.
I may one day enforce this by carrying around two poles that bear "stop", "do not enter + wrong way", "no u-turn", "do not reverse + severe tire damage", and "no stopping any time" signs on them, and creating pop-up paradox traps for unwary driving AIs.
Cities will become completely Undriveable and possibly off limits to cars once the majority are self driving.
Here's why; pedestrians will know that all cars will stop for them and people will just start crossing anytime they want and even start simply walking in the street. There might be some jaywalking enforcement attempts but ultimate they'll fail.
What do you guys think?
With the saturation of safety gear on newer cars you really have no excuse anyhow, the newer ones even slam on the brakes FOR you when they detect an obstacle.
One time someone was standing in the bike lane attempting to cross a 3 lane road against the light. She seemed upset that I was going to ride my bike through the green light anyway, and told me that you have to always stop for pedestrians in the crosswalk regardless of the traffic signal. Unfortunately, pedestrians aren't allowed to enter the crosswalk except when they have a walk signal (that flashing red hand means you're not allowed to cross, though I've never seen a single person follow that rule).
As a cyclist I think we should ban both walking and driving ;)
Well she's right; of course you have to stop for pedestrians in the crosswalk. You can't wilfully hit someone and just say 'my right of way' and be all in the clear.
I think in the UK pedestrians actually always have priority, as long as they are travelling in a straight line and you as a driver are making some kind of turn. So for example a road joining another road at a T junction, pedestrians crossing the joining road have priority over cars, regardless of lights or markings.
Willfully perhaps, but if it's accidental and all other circumstances being equal I would say that's the pedestrian's fault. I don't understand this logic that because the pedestrian is ignorant about how traffic law works that they're exempt from following it, yes the driver should be paying attention at all times and be ready for anything to happen, but if someone breaks a don't walk light and just wanders into traffic, they should be held liable for the resulting accident.
There will most often than not still be people in these driverless cars that we will irritate, and pedestrians that we fear will scorn us. However, I think you make a good point about non-passenger vehicles. Perhaps we will be more likely to walk out in front of an unoccupied Uber that we know will stop for us on an empty street where no one will judge us — but this won't affect the busy streets / scenarios where it's most important that I mention above.
Interesting prediction, nonetheless...
These aren't busy city streets where I'm making that choice, but I'm not getting to the point where I'm being considerate of the driver, I simply don't want to create the situation where I am dependent on their noticing me.
It's a little douchy, but goddamnit, California law says I've got right-of-way in crosswalks. When you're holding a mutex, what you do is get rid of it ASAP and get out of the way.
That's for the safety of the pedestrian in order to avoid confusion and make liability strictly fall on the shoulders of the party with the 2 killing machine.
Regardless of my legal right-of-way, though, I try not to be a jerk about it.
If you need to "slam" on your breaks to let a pedestrian cross, you are driving too fast.
Anyone who has ever lived (or otherwise spent a lot of time driving) in a bad neighborhood knows intimately that this is true. People walk in the street in rough places. And there aren't any norms compelling them to get out of the way.
When you walk up to a doorman, you make eye contact and look for him to open the door. When you walk up to an automatic door, you have been trained to know it will open for you.
Run an experiment with a bad doorman who doesn't open doors, and an automatic door that doesn't open, and you'll have a lot more people running into the automatic door.
Also you'd feel less bad about making a car slow down if you assumed the occupants weren't paying attention and you're just annoying a machine, no?
When a pedestrian is detected on the road outside of a crosswalk, accelerate instead of braking.
> This doesn't happen in countries unlike the US where "jaywalking" is not a crime or even a concept.
I think it should be "This doesn't happen in (countries unlike the US) where "jaywalking" is not a crime of even a concept" rather than how I initially read it "This doesn't happen in countries unlike (the US where "jaywalking is not a crime or even a concept.)"
http://www.citylab.com/commute/2012/04/invention-jaywalking/...
Autonomous vehicles may be even better for this, as a person could be dropped off at the closest accessible point, then the vehicle retreats to a lower-density area.
With that said, I hope they phase out human driving (or require similarly ex(p|t)ensive insurance policies) once SDCs become robust; the joy of driving isn't enough, IMHO to build the entire infrastructure and compromise safety.
My position in this discussion is that yes, infrastructure improvements are necessary, but they'll be relatively straight forward in the US, as basically every city in the US was designed to support automobile traffic.
Self driving cars are going to have a much tougher row to hoe in Europe and Asia, where city infrastructure predates the automobile by centuries/millenia
This is not to say that it's impossible, but US cities barely manage to get the infrastructure upgrades and maintenance they need now - Google et. al. have a lot of money, but I'm not confident they can play bureaucracy well enough to get some of the infrastructure improvements necessary for either large fleets of SDC's. I admit, the idea of replacing all the individually owned cars on the road with a fleet of summonable automobiles is very appealing; it just seems like there are a lot of intermediary steps that are less about technology and money than they are about politics and personal interests.
My favorite article on the subject[1] extrapolates how much impact autonomous vehicle technology will have. I'd actually be pretty surprised if it hadn't come across the desks of the GM executives involved in the half a billion dollar Lyft investment[2]. The Obama administration seems to be making a real effort to lay the groundwork for the necessary infrastructure and regulatory resources[3].
Also, if you haven't read/watched this, don't pass go: http://www.bloomberg.com/features/2015-george-hotz-self-driv...
[1] http://zackkanter.com/2015/01/23/how-ubers-autonomous-cars-w... [2] http://www.bloomberg.com/news/articles/2016-01-04/gm-invests... [3] http://www.usatoday.com/story/money/cars/2016/01/14/nhtsa-de...
I can't believe they are describing this as a custom rather than a dangerous behavior born of impatience. I grew up in a place where people did this and it only took one close call with a clearly drunk driver flying through a red across the intersection in front of me to make me stop for good. Why would anyone design a self driving vehicle to emulate this behavior?
"Why cities aren't ready for the horseless carriage."
... Sure, they may not be. But that doesn't mean that change will stop, or that they have to be ready overnight. It means that we'll have to deal with these things eventually, and I'm betting it's sooner rather than later.
And consider, if you will, how this can become the modern version of "Red Barchetta" if you did choose to keep driving yourself.
I'd rather not ride at all at that point, and that makes me sad as well.
I can't wait for self driving cars for a jillion reasons but I am dismayed when advocates claim that the infrastructure must change to support them. That's completely backwards: our tools should support us, not the other way around.
Self-driving cars in an uncooperative environment is probably an order of magnitude harder than that. Demanding that they work in an uncooperative environment results in a problem that is probably intractably hard.
That's why I'm skeptical of self-driving cars becoming widespread in the U.S. in the next decade. Driving in an unknown environment over millions of miles of roads is going to be a huge challenge if the infrastructure doesn't change, but I can't figure out which political party is going to support the massive infrastructure project. The early adopters of this technology are probably going to be transportation companies. Democrats are going to have a hard time advocating for spending money that throws truck and taxi drivers out of work. And good luck getting Republicans to spend additional money on anything.
I hope that google and/or the car companies pull it off, but I don't think that we should hold our breath waiting for the world to adjust to self-driving cars. The hardware and software will have to do 99% of the work.
1) Traffic will disappear because cars can stop faster and travel close together. At best there are a tiny minority of driving scenarios where driving closer together or stopping quicker are the system bottlenecks. Rush hour traffic is not one of them. At best, you will drive faster on highways, but that will just get you to your exit faster, where you will spend more time waiting than you used to because intracity bottlenecks haven't changed. Traffic lights will still continue to make tradeoffs between latency and throughput, and self driving cars can't take away that tradeoff. And cars will continue to drive slow enough to stop for typical city obstructions (like cyclists, pedestrians, etc.), even if they have a few milliseconds faster of a reaction time. And you're still going to have typical rush hour volumes of people coming from multiple different places converging on a single area.
2) Haven't you seen the simulation where self driving cars don't need traffic signals anymore cause they can drive right through without stopping or hitting anybody? Guess what, it's a simulation...one that conveniently ignored almost every other relevant factor about how the world works. Like pedestrians. Dogs. Joggers. Cyclists. Handicapped people. Gridlocks that happen due to bottlenecks on perpendicular streets. Street protests. Weather conditions. In fact, the only way this idea feasibly works is if you are in a remote intersection in the middle of somewhere so unimportant that it doesn't have any other obstacles or traffic types, in an area where non-driverless cars are banned. And guess what...those situations get along quite fine with traffic lights.
3) But you won't have to circle around the block endlessly to find parking! Actually, your situation isn't much better, and if you see a rise in commuting by cars it could get much worse. Your traffic lanes will be continuously blocked by people getting in and out of their cars. Think of the airport arrivals and departures loops...nobody there has a need for parking, and yet still they are a traffic nightmare because everybody is either getting in a car or getting out of a car, and you have cars blocking traffic looking for their specific passenger, and you have cars circling around the block anyway because there is nowhere to stop legally.
I could go on and on. Traffic is the external manifestation of a system that is far more complex than streets with a few bad drivers clogging shit up. It is a manifestation of the interaction between cars and non-cars, economies and the humans that contribute to them, all with different goals and objectives, all trying to use a shared but very scarce resource. It isn't going away. Ever. The best you can do is find ways to move more people through it faster, and from that perspective the geometric reality of an automobile is grim. No matter how optimized for ommitting the driver it may be, a car will always take up exponential amounts of space compared to public transit systems.
But that isn't a reason to not pursue them. They should still be pursued on safety grounds alone. They just won't answer your prayers for a traffic-free city.