72 comments

[ 3.5 ms ] story [ 129 ms ] thread
I have to turn the pedestrian detection off every time I drive past the local school. Why? The kids figured out that it's really easy to make new cars emergency brake by just pretending to walk onto the road.

People suck.

people suck or kids will be kids?
Same thing I guess? It's humans sabotaging a safety feature intended to protect them.
I think it is more like testing your bounds or figuratively pulling the rope to see if it breaks. Both are innate human (and animal) behavior and in that both a blessing and a curse.
Kids don't know any better yet.

That being said I would probably lose my temper as well when faced with such a situation.

When I'm dropping off my kids and other kids run in my way I usually just stop to chat or play with them, but I'm doing this when walking or on a bicycle.

I guess if I wore a 2-ton suit of power armor I'd just be pissed off instead :)

The main problem here is that the 2-ton suit of power armor doesn't really care about what's behind it when it slams the brakes for a pedestrian, you would probably be pretty upset if you were biking behind me when this happens.

I think it's a great thing that the car prioritizes pedestrian safety, but it's very unfortunate that people would abuse this.

If they don't know by that age (attending school), something is very wrong.
Why are you excusing shitty behavior with a trite phrase?
Cars are the leading cause of dead children in the US and this is what annoys you?
How'd you arrive at this conclusion? That's a bizarrely uncharitable interpretation of what I wrote.

I assure you, I'm annoyed by many other things too.

>Cars are the leading cause of dead children

You know, this obviously is exactly why it annoys me to turn off the pedestrian detection. It's not like it protects me, sitting in my car.

it also protects you in the car, if not from physical damage which at speed is likely when hitting a heavy object like a human, from mental trauma of killing a person...
The idea behind what I wrote (which I realize now was unpopular) was that a somewhat overzealous safety system shouldn't be a big deal compared to the possibility of it preventing thousands of dead kids a year, if deployed at scale. Asking such a system to distinguish pretend behavior seems like a big ask.
>somewhat overzealous safety system shouldn't be a big deal compared to the possibility of it preventing thousands of dead kids a year

I totally agree with you on this, I was just observing an unfortunate byproduct of an otherwise great system.

>Asking such a system to distinguish pretend behavior seems like a big ask.

I agree, I'm certainly not demanding that anyone do anything to "solve" this issue.

lol, cars arent the cause. people who can't drive are.
Truth be told I personally think infrastructure and cultural norms that take automobiles as the default preferred road use are the real cause, but at the point of impact the car, operated by a driver (or machine) is the cause.
Personally, I would also brake if a kid pretended to walk into the road. There just has to be one kid who isn't pretending for it to become a really big issue.
As a human, this behavior does not look like a kid walking into the road. They wait, take a single exaggerated step forwards and immediately back off.

But hey, thanks for reminding me to buy a dashcam.

Beginning of pranks to come with self drive cars.

Imagine when someone realised you can slam on the brakes in front of X brand of car and it prioritises avoiding the car in front over all else

Hopefully the expensive self driving car can afford to integrate a $40 dashcam.
(comment deleted)
You should be a more defensive driver. This sort of attitude could lead to you seriously hurting someone.
What sort of attitude? Could you be more specific? Turning the pedestrian detection off in that specific area is defensive driving, I don't want to get rear-ended because someone maliciously triggers it.

I live in an extremely busy city full of scooters (of all kinds!) and drunk pedestrians, it's not possible to get far here without extreme care.

I’m a fan of Tesla but when you see that their system is not a bit better than the competition (they’re all failing in low light conditions), I don’t know how they can claim autonomous driving “coming soon.” Especially after Uber’s death you’d think that this specific condition would be given extremely high priority.
They recognize much more objects on the road but major problem is to do not cause a lot of false alarms. e.g. You don't really want to drive car which decided that plastic bag on the highway is reason for emergency braking, on the other hand is that really plastic bag or some drunk person who is sleeping on the road...

I remember when Volvo introduced IntelliSafe technology few years ago that car was totally "paranoid" and it was beeping all the time -> most of drivers deactivated that system.

If it's confused between a plastic bag and a human, it hasn't recognised it.
It didn't recognize a plastic bag as a human, it recognized a plastic bag as an unknown object. A solid object the size of a plastic bag could cause a major accident.
How good are humans detecting pedestrians in low light conditions?

It's an important test case, but I wonder how many of those would be undetectable by a human as well.

Ideally some 'low light camera' should warn you that in darkness far ahead there's 'something' unusual on the road and you will have time to react. Recently there was accident where truck with palm oil overturned on highway and first cars crashed into the truck because they were travelling too fast and only with low beam lights : https://www.youtube.com/watch?v=igX765Vw1fY

For example this camera module will be used soon in cars and it's really more sensitive than driver's eye : http://www.camera-module.com/download/sony-imx390-sensor-sel...

Adding thermal imaging to the sensor suite would probably help with a lot of edge cases, but the hardware is still very expensive. It has come down some in the last 5-10 years but there is a long way to go.
The necessary sensors are really expensive, and that's why the production cars from Tesla and GM and such just aren't that great at it.
> they’re all failing in low light conditions

Isn't this an issue with lights ?

No, because car lights only illuminate some of the surrounding area. You can't illuminate much more without causing issues for other road users (which is why there are laws around car head lamp & bulb design & usage).
Easy - just remove windows on cars so other car users are not impacted.
Having actually test-driven many of these cars this year, in practice Tesla is much better than the competition. There might still be problems in different conditions, and it is not fully autonomous of course, but the common HN myth that other cars basically have the same lane assist that tesla has is just a lore. Tesla autopilot is light years ahead of the rest of them.
Look at the IIHS ratings though. The Tesla failed to detect a crossing child in both the 12 and 25mph tests. In both, it still hit the child with only a 5mph speed reduction.

Seems unlike Tesla to be out-classed like this. Six other models had significantly better speed reduction or didn't hit the pedestrian at all.

https://www.iihs.org/news/detail/performance-of-pedestrian-c...

This is definitely interesting, and it is completely possible that in some situations other manufacturers might have the upper hand.

They did test it on autopilot though, right? It says "standard" on Model 3. AFAIK autopilot features are not standard, it's an option you have to buy on Tesla. I also can't find anywhere in the report whether the car was actually driving on autopilot or not. All of these details I think make a big difference. A standard non-autopilot-equipped model 3 might just be subpar to other cars, it would not surprise me.

>I don’t know how they can claim autonomous driving “coming soon.”

How is easy: it doesn't cost anything, and makes them profit.

It's like how the had the gal to nam their system "Autopilot" when it's not, and full awareness is required from the driver in the small print...

I own a 3 and my opinion is while people are raving over the new summon features they should instead realize that it also shows how far off autonomous driving truly is. Just go watch all the videos, all we truly have is a 60k RC car that moves slower than a RC car and still requires permission to move.

Now the current software does very well on limited access roads; highways and interstates; and has for over a year that I have owned the car. It does fine on two lane roads but the use cases there are far more limited and mostly to showing off for friends. Well that and doing stuff you normally would wait till your stopped to do.

Basically none of them are good enough for a real world scenario in a busy city. But they may save life on some scenarios which is good.

They can achieve a huge safety improvement by forcing speed limit. But somehow car manufacturers don’t want to implement it.

> They can achieve a huge safety improvement by forcing speed limit. But somehow car manufacturers don’t want to implement it.

I've had multiple cars that show the current speed limit on the dash. Some based on GPS, others based on visual recognition. None of them have been correct 100% of the time. I wouldn't want my car to slow me down on a highway because it misinterprets a sign or because their data isn't up-to-date, or because the GPS is glitching and thinks I'm driving on a side road.

It's a really simple solution in theory, but in practice it doesn't seem feasible yet.

> But somehow car manufacturers don’t want to implement it.

Maybe because nobody would buy a car that enforces its idea of the current speed limit on the driver? (It'd be good to have as an option though, I agree, as long as it can be overridden.)

> Basically none of them are good enough for a real world scenario in a busy city

So what's "good enough"? Even if these systems don't prevent all collisions they'll often vastly reduce the speeds at which they happen.

Not exactlty the same thing, but Volvo has stated they plan to limit the top speed of all their cars to 180km/h
I think the interesting thing in that article is the stat that pedestrian fatalities reached their lowest in 2009, and have been increasing since. Is it really (as I've heard and read) that it's because people are on their phones and walking out into the road? I've definitely seen people do this in front of me, but I know that's merely anecdotal.
And at the same time people are also on their phones in their cars, decreasing the chance that they can respond to that person walking into the road. I've had so many times where I didn't have an accident just because I was paying attention to what other people were doing.
> "in most cases avoided hitting the pedestrian"

I am not sure if this result merits the use of the word "aced" in the title.

I think it's fair to say that whether you hit the pedestrian is more or less irrelevant to your score on a pedestrian detection test.
Well, it's not irrelevant to a "timely pedestrian detection" test, which is what actually matters in self-driving cars...
the Audi A4, BMW 3 series, Subaru Outback, Mercedes-Benz C-Class, Nissan Maxima, and Volvo S60.
With the news about Uber and Tesla accidents, it’s been making me wonder: why not equip these cars with one more sensor: an infrared camera. Combining thermal information with extant information will be a great boon in making intelligent decisions, no?
I 100% agree but I would bet the issue is unit cost for the hardware. Anything with reasonable resolution is going to be in the four digit dollar range per sensor.
This seems totally reasonable for an actual damn autopilot
$250 gets you 206x156 today
Interesting that Tesla, with what I would have assumed to be a much more sophisticated system didn’t do as well.

Anyone care to have a guess why?

The pedestrian detection and automatic emergency braking systems are separate from autopilot. They are always on, even if autopilot is not in use. As I understand it, they are different code, more conventional and similar to the systems in other cars, where autopilot is a grand research project/extended beta. As with most things Tesla they are likely to improve over time with OTA updates.
Virtually eliminating pedestrian fatalities could be quite simple, but we don't do it because it would inconvenience people: Set a 30 km/h (18 mph) speed limit everywhere but controlled-access freeways, and enforce it strictly. At 30km/h a car-pedestrian collision is almost always non-fatal[1], and at that speed cars can stop almost instantaneously.

1: https://nacto.org/docs/usdg/relationship_between_speed_risk_...

The good news is that cities are starting to finally do this. Slowly, but it is happening.
That means halving the bandwidth of all inner-city roads (that are traditionally at 35mph+)

If a road used to carry 1000-cars per hour, it will only carry 500-cars per hour if you drop the speed limit from 35 -> 18. Effectively doubling the length of time for congestion.

You can get around this by building twice as many roads I guess, but cities don't really have any more room.

If there are 2 seconds of distance between cars at 36mph and 2 seconds of distance between cars at 18mph then the road has exactly the same capacity at both speeds.

Also, the constraint on street capacity is intersections in the vast majority of cases. Until you remove intersections nothing else really else matters.

I don't think people are "timing" their cars, and are instead driving at "car-lengths" worth of distance between the person in front of them

If you instead assume 1-car lengths worth of distance at both 36mph and 18mph, the road capacity shrinks dramatically between the two speeds.

If that is true, most congestion on my way to work is after an accident. Not having that accident at 30kph is much more likely than not having it at twice the speed, giving the same distance to react and brake.
You're also underselling the intersections part, because getting traffic to intermix at intersections is incredibly easy at 20mph. And, what you've conceptually stated (no loss in capacity) has been empirically demonstrated.

Finally, you get vastly different places at 40 mph vs 20mph. At 20, roads are smaller, people can safely exist close to the roadway and even enjoy themselves, and you don't need huge buffer zones for the safety of drivers or people. All these things compound to put people (willingly) in closer proximity to one another, so that, while peak travel speed is lower, traffic throughput stays relatively consistent, and travel times decrease overall because many long trips get replaced with short ones.

I'm betting that if pedestrians think they are safer, they will interrupt the flow of traffic much more often, which will impact throughput.
artificially setting limits like that is a poor solution, especially when you're depending on police enforcement to ensure compliance.

a better solution is narrowing lanes (to 9-10 ft), and possibly installing bollards everywhere, because that internalizes the costs of collision (car gets wrecked and you get hurt in addition to pedestrians), so drivers naturally self-regulate driving behavior for safety (rather than the secondary measure, speed). you also don't need an external police force and it's attendant costs, both financially and societally, for enforcement.

How about a device that pedestrians could optionally carry that would broadcast their presence to all vehicles