In the piece — titled “Can You Fool a Self Driving Car?” — Rober found that a Tesla car on Autopilot was fooled by a Wile E. Coyote-style wall painted to look like the road ahead of it, with the electric vehicle plowing right through it instead of stopping.
The footage was damning enough, with slow-motion clips showing the car not only crashing through the styrofoam wall but also a mannequin of a child. The Tesla was also fooled by simulated rain and fog.
E. Lon Musk. Supah. Geenius.
MEEP MEEP
I hope some of you actually skimmed the article and got to the “disengaging” part.
As Electrek points out, Autopilot has a well-documented tendency to disengage right before a crash. Regulators have previously found that the advanced driver assistance software shuts off a fraction of a second before making impact.
It’s a highly questionable approach that has raised concerns over Tesla trying to evade guilt by automatically turning off any possibly incriminating driver assistance features before a crash.
Don’t get me wrong, autopilot turning itself off right before a crash is sus and I wouldn’t put it past Tesla to do something like that (I mean come on, why don’t they use lidar) but maybe it’s so the car doesn’t try to power the wheels or something after impact which could potentially worsen the event.
On the other hand, they’re POS cars and the autopilot probably just shuts off cause of poor assembly, standards, and design resulting from cutting corners.
Wouldn’t it make more sense for autopilot to brake and try to stop the car instead of just turning off and letting the car roll? If it’s certain enough that there will be an accident, just applying the brakes until there’s user override would make much more sense…
Rober seems to think so, since he says in the video that it’s likely disengaging because the parking sensors detect that it’s parked because of the object in front, and it shuts off the cruise control.
Normal cars do whatever is in their power to cease movement while facing upright. In a wreck, the safest state for a car is to cease moving.
I see your point, and it makes sense, but I would be very surprised if Tesla did this. I think the best option would be to turn off the features once an impact is detected. It shutting off before hand feels like a cheap ploy to avoid guilt
It always is that way; fuck the consumer, its all about making a buck
I’ve heard that too, and I don’t doubt it, but watching Mark Rober’s video, it seems like he’s deathgripping the wheel pretty hard before the impact which seems more likely to be disengaging. Each time, you can see the wheel tug slightly to the left, but his deathgrip pulls it back to the right.
deleted by creator
To be fair, if you were to construct a wall and paint it exactly like the road, people will run into it as well. That being said, tesla shouldn’t rely on cameras
To be fair, if you were to construct a wall and paint it exactly like the road, people will run into it as well.
this isn’t being fair. It’s being compared to the other- better- autopilot systems that use both LIDAR and radar in addition to daylight and infrared optical to sense the world around them.
Teslas only use daylight and infrared. LIDAR and radar systems both would not have been deceived.
The video does bring up human ability too with the fog test (“Optically, with my own eyes, I can no longer see there’s a kid through this fog. The lidar has no issue.”) But, as they show, this wall is extremely obvious to the driver.
The tesla would lose its shit if it sees this
They already have trouble enough with trucks carrying traffic lights, or with speed limit drivers on them.
Yeah, the Roadrunner could easily skip by such barriers, frustrating the Coyote to no end. Tesla is not a Roadrunner.
It’s a highly questionable approach that has raised concerns over Tesla trying to evade guilt by automatically turning off any possibly incriminating driver assistance features before a crash.
So, who’s the YouTuber that’s gonna test this out? Since Elmo has pushed his way into the government in order to quash any investigation into it.
It basically already happened in the Mark Rober video, it turns off by itself less than a second before hitting
My 500$ robot vacuum has LiDAR, meanwhile these 50k pieces of shit don’t 😂
Holy shit, I knew I’d heard this word before. My Chinese robot vacuum cleaner has more technology than a tesla hahahahaha
Vacuum doesn’t run outdoors and accidentally running into a wall doesn’t generate lawsuits.
But, yes, any self-driving cars should absolutely be required to have lidar. I don’t think you could find any professional in the field that would argue that lidar is the proper tool for this.
…what is your point here, exactly? The stakes might be lower for a vacuum cleaner, sure, but lidar - or a similar time-of-flight system - is the only consistent way of mapping environmental geometry. It doesn’t matter if that’s a dining room full of tables and chairs, or a pedestrian crossing full of children.
I think you’re suffering from not knowing what you don’t know.
Let me make it a but clearer for you to make a fair answer.
Take a .25mw lidar sensor off a vacuum, take it outdoors and scan an intersection.
Will that laser be visible to the sensor?
is it spinning fast enough to track a kid moving in to an intersection when you’re traveling at 73 feet per second?
You’re mischaracterizing their point. Nobody is saying take the exact piece of equipment, put it in the vehicle and PRESTO. That’d be like asking why the vacuum battery can’t power the car. Because duh.
The point is if such a novelty, inconsequential item that doesn’t have any kind of life safety requirements can employ a class of technology that would prevent adverse effects, why the fuck doesn’t the vehicle? This is a design flaw of Teslas, pure and simple.
But they do, there are literally cars out there with lidar sensors.
The question was why can’t I have a lidar sensor on my car if my $150 vacuum has one. The lidar sensor for a car is more than $150.
You don’t have one because there are expensive at that size and update frequency. Sensors that are capable of outdoor mapping at high speed cost the price of a small car.
The manufacturers suspect and probably rightfully so that people don’t want to pay an extra 10 - 30 grand for an array of sensors.
The technology readily exists rober had one in his video that he used to scan a roller coaster. It’s not some conspiracy that you don’t have it on cars and it’s not like it’s not capable of being done because waymo does it all the time.
There’s a reason why waymo doesn’t use smaller sensors they use the minimum of what works well. Which is expensive, which people looking at a mid-range car don’t want to take on the extra cost, hence it’s not available
Only Tesla does not use radar with their control systems. Every single other manufacturer uses radar control mixed with the camera system. The Tesla system is garbage.
The self driving system uber was working on also went downhill after they went full visual only.
yeah, you’d think they’d at least use radar. That’s cheap AF. It’s like someone there said I have this hill to die on, I bet we can do it all with cameras.
I think you’re suffering from not knowing what you don’t know.
and I think you’re suffering from being an arrogant sack of dicks who doesn’t like being called out on their poor communication skills and, through either a lack of self-awareness or an unwarranted overabundance of self-confidence, projects their own flaws on others. But for the more receptive types who want to learn more, here’s Syed Saad ul Hassan’s very well-written 2022 paper on practical applications, titled Lidar Sensor in Autonomous Vehicles which I found also serves as neat primer of lidar in general..
Well look at you being adult and using big words instead of just insulting people. Not even going to wastime on people like you, I’m going to block you and move on and hope that everyone else does the same so you can sit in your own quiet little world wondering why no one likes you.
You’re an idiot.
jesus man, how many alts do you have?
And the president is driving one of these?
Maybe we should be purchasing lots of paint and cement blockades…
When he was in the Tesla asking if he should go for a ride I was screaming “Yes! Yes Mr. President! Please! Elon, show him full self driving on the interstate! Show him full self driving mode!”
The president can’t drive by law unless on the grounds of the White House and maybe Camp David. At least while in office. They might be allowed to drive after leaving office…
This isn’t true at all. I can’t tell if you’re being serious or incredibly sarcastic, though.
The reason presidents (and generally ex presidents, too) don’t drive themselves is because the kind of driving to escape an assassination attempt is a higher level of driving and training than what the vast majority of people ever have. There’s no law saying presidents are forbidden from driving.
In any case, I would be perfectly happy if they let him drive a CT and it caught fire. I’d do a little jib, and I wouldn’t care who sees that.
Current and past presidents are prohibited from driving.
you’re gonna have to drop a source for that.
because, no, they’re not. the Secret Service provides a driver specially trained for the risks a president might face, and very strongly insists, but they’re not “prohibited” from driving simply because they’re presidents.
to be clear, the secret service cannot prohibit the president from doing anything they really want to do. Even if it’s totally stupid for them to do that. (This includes, for example, Trump’s routine weekend round of golf at Turd-o-Lardo)
I don’t think Trump can drive. As in, he doesn’t even know what the pedals do.
clearly knows what he is doing
He looks like he’s making the siren sounds and having a great time
He’s going to fall out of the cab on the next right turn.
Are his hands even big enough to hold the wheel?
The real question is, in a truly self-driving car, (not a tesla) are you actually driving?
Dang
If you own a tesla or a cybertruck you deserve it.
deleted by creator
It was super annoying how scared he acted when he knew it was styrofoam and it wasn’t even going to leave a scratch on the car. I would have like it much better if the car crashed into and actual wall and burst into flames.
Instinctively, human brains generally don’t like large objects coming to them unbidden at high speed. That isn’t going to help things, even if you’re consciously aware that the wall is relatively harmless.
Looney Tunes shit.
I keep praying that an anvil falls on Elon Husk next.
Okay, that is fucking funny. 😂
If you get any strong emotions on material shit when someone makes a video…you have 0 of my respect. Period.
Saw a guy smash a Stradivarius on video once. definitely had strong emotions on that one.
Really torn up about not having your respect tho…
I think you could argue that that’s not just material stuff though. That’s historical and significant culturally.
Idk if the video has reason to embue strong emotions then it’s fair
Lol yeah they’re “furious”
I have no clue what you’re trying to say, but the significant amount of outrage a day or two later that I suddenly saw explode on Twitter was mind boggling to me. Couldn’t tell if it was bots or morons but either way, people are big mad about the video.
“Dipshit Nazis mad at facts bursting their bubble is unreality” is another way of reading this headline.
well yeah, happens every time I say something mad about their current favorite GPU-use fad.
I believe the outrage is that the video showed that autopilot was off when they crashed into the wall. That’s what the red circle in the thumbnail is highlighting. The whole thing apparently being a setup for views like Top Gear faking the Model S breaking down.
Autopilot shuts itself off just before a crash so Tesla can deny liability. It’s been observed in many real-world accidents before this. Others have said much the same, with sources, in this very thread.
well yes but as long as there’s deniability built into my toy, then YOU’RE JUST A BIG DUMB MEANIE-PANTS WHO HATES MY COOL TOYS BECAUSE YOU DON’T HAVE ONE because there’s no other possible reason to hate a toy this cool.
As Electrek points out, Autopilot has a well-documented tendency to disengage right before a crash. Regulators have previously found that the advanced driver assistance software shuts off a fraction of a second before making impact.
This has been known.
They do it so they can evade liability for the crash.
The self-driving equivalent of “Jesus take the wheel!”
That makes so little sense… It detects it’s about to crash then gives up and lets you sort it?
That’s like the opposite of my Audi who does detect I’m about to hit something and gives me either a warning or just actively hits the brakes if I don’t have time to handle it.
If this is true, this is so fucking evil it’s kinda amazing it could have reached anywhere near prod.The point is that they can say “Autopilot wasn’t active during the crash.” They can leave out that autopilot was active right up until the moment before, or that autopilot directly contributed to it. They’re just purely leaning into the technical truth that it wasn’t on during the crash. Whether it’s a courtroom defense or their own next published set of data, “Autopilot was not active during any recorded Tesla crashes.”
even your audi is going to dump to human control if it can’t figure out what the appropriate response is. Granted, your Audi is probably smart enough to be like “yeah don’t hit the fucking wall,” but eh… it was put together by people that actually know what they’re doing, and care about safety.
Tesla isn’t doing this for safety or because it’s the best response. The cars are doing this because they don’t want to pay out for wrongful death lawsuits.
If this is true, this is so fucking evil it’s kinda amazing it could have reached anywhere near prod.
It’s musk. he’s fucking vile, and this isn’t even close to the worst thing he’s doing. or has done.
Any crash within 10s of a disengagement counts as it being on so you can’t just do this.
Edit: added the time unit.
Edit2: it’s actually 30s not 10s. See below.
Where are you seeing that?
There’s nothing I’m seeing as a matter of law or regulation.
In any case liability (especially civil liability) is an absolute bitch. It’s incredibly messy and likely will not every be so cut and dry.
Well it’s not that it was a crash caused by a level 2 system, but that they’ll investigate it.
So you can’t hide the crash by disengaging it just before.
Looks like it’s actually 30s seconds not 10s, or maybe it was 10s once upon a time and they changed it to 30?
The General Order requires that reporting entities file incident reports for crashes involving ADS-equipped vehicles that occur on publicly accessible roads in the United States and its territories. Crashes involving an ADS-equipped vehicle are reportable if the ADS was in use at any time within 30 seconds of the crash and the crash resulted in property damage or injury
https://www.nhtsa.gov/sites/nhtsa.gov/files/2022-06/ADAS-L2-SGO-Report-June-2022.pdf
Thanks for that.
The thing is, though the NHTSA generally doesn’t make a determination on criminal or civil liability. They’ll make the report about what happened and keep it to the facts, and let the courts sort it out whose at fault. they might not even actually investigate a crash unless it comes to it. It’s just saying “when your car crashes, you need to tell us about it.” and they kinda assume they comply.
Which, Tesla doesn’t want to comply, and is one of the reasons Musk/DOGE is going after them.
I knew they wouldn’t necessarily investigate it, that’s always their discretion, but I had no idea there was no actual bite to the rule if they didn’t comply. That’s stupid.
10n what
Oops haha, 10 seconds.
If it knows it’s about to crash, then why not just brake?
So, as others have said, it takes time to brake. But also, generally speaking autonomous cars are programmed to dump control back to the human if there’s a situation it can’t see an ‘appropriate’ response to.
what’s happening here is the ‘oh shit, there’s no action that can stop the crash’, because braking takes time (hell, even coming to that decision takes time, activating the whoseitwhatsits that activate the brakes takes time.) the normal thought is, if there’s something it can’t figure out on it’s own, it’s best to let the human take over. It’s supposed to make that decision well before, though.
However, as for why tesla is doing that when there’s not enough time to actually take control?
It’s because liability is a bitch. Given how many teslas are on the road, even a single ruling of “yup it was tesla’s fault” is going to start creating precedent, and that gets very expensive, very fast. especially for something that can’t really be fixed.
for some technical perspective, I pulled up the frame rates on the camera system (I’m not seeing frame rate on the cabin camera specifically, but it seems to either be 36 in older models or 24 in newer.)
14 frames @ 24 fps is about 0.6 seconds@36 fps, it’s about 0.4 seconds. For comparison, average human reaction to just see a change and click a mouse is about .3 seconds. If you add in needing to assess situation… that’s going to be significantly more time.
AEB braking was originally designed to not prevent a crash, but to slow the car when a unavoidable crash was detected.
It’s since gotten better and can also prevent crashes now, but slowing the speed of the crash was the original important piece. It’s a lot easier to predict an unavoidable crash, than to detect a potential crash and stop in time.
Insurance companies offer a discount for having any type of AEB as even just slowing will reduce damages and their cost out of pocket.
Not all AEB systems are created equal though.
Maybe disengaging AP if an unavoidable crash is detected triggers the AEB system? Like maybe for AEB to take over which should always be running, AP has to be off?
Because even braking can’t avoid the crash. Unavoidable crash means bad juju if the ‘self driving’ car image is meant to stick around.
Breaks require a sufficient stopping distance given the current speed, driving surface conditions, tire condition, and the amount of momentum at play. This is why trains can’t stop quickly despite having breaks (and very good ones at that, with air breaks on every wheel) as there’s so much momentum at play.
If autopilot is being criticized for disengaging immediately before the crash, it’s pretty safe to assume its too late to stop the vehicle and avoid the collision
This autopilot shit needs regulated audit log in a black box, like what planes or ships have.
In no way should this kind of manipulation be legal.
Not sure how that helps in evading liability.
Every Tesla driver would need super human reaction speeds to respond in 17 frames, 680ms(I didn’t check the recording framerate, but 25fps is the slowest reasonable), less than a second.
They’re talking about avoiding legal liability, not about actually doing the right thing. And of course you can see how it would help them avoid legal liability. The lawyers will walk into court and honestly say that at the time of the accident the human driver was in control of the vehicle.
And then that creates a discussion about how much time the human driver has to have in order to actually solve the problem, or gray areas about who exactly controls what when, and it complicates the situation enough where maybe Tesla can pay less money for the deaths that they are obviously responsible for.
They’re talking about avoiding legal liability, not about actually doing the right thing. And of course you can see how it would help them avoid legal liability. The lawyers will walk into court and honestly say that at the time of the accident the human driver was in control of the vehicle.
The plaintiff’s lawyers would say, the autopilot was engaged, made the decision to run into the wall, and turned off 0.1 seconds before impact. Liability is not going disappear when there were 4.9 seconds of making dangerous decisions and peacing out in the last 0.1.
The plaintiff’s lawyers would say, the autopilot was engaged, made the decision to run into the wall, and turned off 0.1 seconds before impact. Liability is not going disappear when there were 4.9 seconds of making dangerous decisions and peacing out in the last 0.1.
these strategies aren’t about actually winning the argument, it’s about making it excessively expensive to have the argument in the first place. Every motion requires a response by the counterparty, which requires billable time from the counterparty’s lawyers, and delays the trial. it’s just another variation on “defend, depose, deny”.
They can also claim with a straight face that autopilot has a crash rate that is artificially lowered without it being technically a lie in public, in ads, etc
Defense lawyers can make a lot of hay with details like that. Nothing that gets the lawsuit dismissed but turning the question into “how much is each party responsible” when it was previously “Tesla drove me into a wall” can help reduce settlement amounts (as these things rarely go to trial).
Which side has more money for lawyers though?
It’s not likely to work, but them swapping to human control after it determined a crash is going to happen isn’t accidental.
Anything they can do to mire the proceedings they will do. It’s like how corporations file stupid junk motions to force plaintiffs to give up.
deleted by creator
If the disengage to avoid legal consequences feature does exist, then you would think there would be some false positive incidences where it turns off for no apparent reason. I found some with a search, which are attributed to bad software. Owners are discussing new patches fixing some problems and introducing new ones. None of the incidences caused an accident, so maybe the owners never hit the malicious code.
if it randomly turns off for unapparent reasons, people are going to be like ‘oh that’s weird’ and leave it at that. Tesla certainly isn’t going to admit that their code is malicious like that. at least not until the FBI is digging through their memos to show it was. and maybe not even then.
I think Mark (who made the OG video) speculated it might be the ultrasonic parking sensors detecting something and disengaging.
That does sound more reasonable.
