Welcome to Tesla Motors Club
Discuss Tesla's Model S, Model 3, Model X, Model Y, Cybertruck, Roadster and More.
Register

Autopilot: Crashed at 40mph

This site may earn commission on affiliate links.
This is a completely irresponsible way to design "automatic emergency braking".

In railroad trains, the rule is that every train must maintain an absolute safe distance behind the train in front of it -- it must be able to stop before reaching the train in front of it. If the train in front stops dead on a dime (perhaps because a bridge fell on it or it hit a boulder from a rockslide), the train behind it will hit the brakes and stop before colliding with anything. The automatic braking systems -- which are now mandated! -- do exactly this.

This is the rule I follow when driving. Always assume that the car ahead of you could stop completely still at any time and be prepared to brake to a dead stop without hitting it. This saved my life once, driving on an HOV lane in California with concrete barriers on each side at 70 mph, when a turn revealed a car which was completely stopped. I slammed on the brakes and stopped ~1000 feet behind it. If I had been tailgating like *almost everyone else on the road*, there would have been a nasty crash.


This is Tesla's fault. The TACC is not driving safely. Sure, most drivers don't drive safely *either*, but for guys like me and Canuck, the Autopilot is simply worse than a human. This is not a good design.

Well, if it is Tesla's fault then it is also the fault of a lot of other auto makers. There seems to be a fear among auto makers that a false positive might cause their car to brake very hard and cause an accident. Hence some MB systems that only apply 40% braking. I say some since I am not familiar with all MB systems and don't want to make a blanket statement.

Despite these limitations the systems reduce the number of accidents and the severity of those that do happen. Not one system I know of avoids all accidents. Here is what the IIHS says:

Systems with automatic braking reduce rear-end crashes by about 40 percent on average, while forward collision warning alone cuts them by 23 percent, the study found. The autobrake systems also greatly reduce injury crashes.

Notice that the systems do not eliminate rear end accidents.
 
Well, if it is Tesla's fault then it is also the fault of a lot of other auto makers.
Agreed. Absolutely correct.

There seems to be a fear among auto makers that a false positive might cause their car to brake very hard and cause an accident. Hence some MB systems that only apply 40% braking.


Notice that the systems do not eliminate rear end accidents.
No safety system can stop someone *without* the system from rear-ending *you*.

Liability in a rear-ender always attaches to the car in the back who crashed head-on. The person in front is practically always blameless legally, at least in my state. I want my car's safety systems to understand this and prevent me from hitting anyone; anyone hitting me, it's their problem.

The only way to eliminate rear-end crashes is (a) to have a system which makes sure that *your* car never rear-ends anyone *else*, and (b) to require every car to have it. This is what they are doing for railroad trains.
 
Agreed. Absolutely correct.





No safety system can stop someone *without* the system from rear-ending *you*.

Liability in a rear-ender always attaches to the car in the back who crashed head-on. The person in front is practically always blameless legally, at least in my state. I want my car's safety systems to understand this and prevent me from hitting anyone; anyone hitting me, it's their problem.

The only way to eliminate rear-end crashes is (a) to have a system which makes sure that *your* car never rear-ends anyone *else*, and (b) to require every car to have it. This is what they are doing for railroad trains.

I fully understand your point of view. I just found the 40% number in the MB manual interesting. I think as more cars have the systems and as manufacturers get more experience with the systems that we will see more aggressive braking. Mots of the current systems were designed for a single camera (not stereoscopic) at best.
 
  • Like
Reactions: neroden
as the brand under discussion comment, if people criticize Tesla for something and aren't willing to criticize the same issue on another brand then I wonder what their agenda is. If Tesla is inferior then that is a different issue.
I have no agenda other than trying to educate other Tesla (would be) owners. I have no reason to criticize any other brand because I didn't just spend €120.000 on one of their cars, no, I bought a Tesla instead. So my agenda is with Tesla. I couldn't give less what other brands say or do, I care about my car and its maker, and others driving the same car.

Putting down other brands to relativate and downplay issues with Tesla don't make the issues disappear. Even if every car manufacturer had this and they all failed in a similar fashion, my agenda as you called it would only be with the brand I chose to go with, and I wouldn't give a [+${$~%{>] about other brands.
 
  • Like
Reactions: KZKZ and X Yes?
This is Tesla's fault. The TACC is not driving safely. Sure, most drivers don't drive safely *either*, but for guys like me and Canuck, the Autopilot is simply worse than a human. This is not a good design.

Tesla Autopilot is great in certain robotic predictable tasks but it is much worse or fatal if you expose it to unpredictable scenario without a sound driver judgement.

Just because Tesla is here it doesn't mean Autopilots would stop airplane crashes and automobile accidents.

A good autopilot design would make sure airplanes would never crash and car would never collide.

We are not at that point yet.

People still die but at least by introducing these imperfect technology incrementally as we know how and as we get better, there have been fewer deaths.
 
You can only do so much with the available resources in in hands.

The system will be upgraded and advance incrementally so in the mean time, read your manual and aware of its limitations.
True as that may be, this sort of info belongs in blog posts, videos, and DS briefings in my opinion. Not just the manual. I read the manual 5 times before even getting the car yet was lulled by the AP systems into believing they could handle anything when in fact they couldn't. I learned the hard way that the manual wasn't exaggerating and what to do in retrospect in certain situations. In the three months since my accident and recommendations to Tesla to please educate other owners more on this subject, all they've done is cover themselves in incident after incident by either saying the AP was off or worked as intended. But to simply put out a video demonstrating the (in)capabilities and proper coping procedures was apparently deemed to be bad marketing or not worth their time. When in fact it is a known case to them and one they frequently test on, as was recently made evident by a video shot from a helicopter showing a Model S approaching an inflatable rear end of a car and doing some AEB tests. Why not use that testing equipment to film an advisory video showing the shortcomings and how to deal with them?

No, just the manuals won't do. Trust me, I know from experience.
 
I have no agenda other than trying to educate other Tesla (would be) owners. I have no reason to criticize any other brand because I didn't just spend €120.000 on one of their cars, no, I bought a Tesla instead. So my agenda is with Tesla. I couldn't give less what other brands say or do, I care about my car and its maker, and others driving the same car.

Putting down other brands to relativate and downplay issues with Tesla don't make the issues disappear. Even if every car manufacturer had this and they all failed in a similar fashion, my agenda as you called it would only be with the brand I chose to go with, and I wouldn't give a [+${$~%{>] about other brands.

Thanks for the reply. I do see your point. As an engineer my viewpoint is to look at what other manufacturers are able to accomplish and use that as a guide as to what is reasonable to expect from the technology. I do expect the systems to progress and I expect that progression to be fairly rapid. We have entered an era where the electronics and, more importantly, the software are evolving rapidly and becoming more and more of what the car is about.
 
  • Like
Reactions: Ivo-G
This is a completely irresponsible way to design "automatic emergency braking".

In railroad trains, the rule is that every train must maintain an absolute safe distance behind the train in front of it -- it must be able to stop before reaching the train in front of it. If the train in front stops dead on a dime (perhaps because a bridge fell on it or it hit a boulder from a rockslide), the train behind it will hit the brakes and stop before colliding with anything. The automatic braking systems -- which are now mandated! -- do exactly this.

This is the rule I follow when driving. Always assume that the car ahead of you could stop completely still at any time and be prepared to brake to a dead stop without hitting it. This saved my life once, driving on an HOV lane in California with concrete barriers on each side at 70 mph, when a turn revealed a car which was completely stopped. I slammed on the brakes and stopped ~1000 feet behind it. If I had been tailgating like *almost everyone else on the road*, there would have been a nasty crash.


This is Tesla's fault. The TACC is not driving safely. Sure, most drivers don't drive safely *either*, but for guys like me and Canuck, the Autopilot is simply worse than a human. This is not a good design.
My 2010 Prius had AEB that was designed exactly the same way. As far a i know, most AEB systems are designed this way, for good reason. The cost of a false positive (slamming on the brakes in heavy traffic) is extremely high from a liability perspective. On the other hand, anything done to reduce the severity of a crash is a net plus. So, reducing the severity of a frontal impact while mediating the severity of a false positive (by not doing panic braking to zero speed) is a reasonable engineering trade off.
 
  • Informative
  • Like
Reactions: bhzmark and neroden
To ban autopilot, would be to ban auto steer. If that were to occur, wouldn't one have to ban any form of "lane keeping" provided by other companies ? I just have a few more questions: would you read a manual fm your life depended on it ? Most people using these feature learned enough to double click the lower stalk, but selectively ignored everything else. Is it OK as long as you can blame your crash on someone else ? Can't always do that. Get used to it. What ever happened to being accountable for your own actions.

Note to Tesla: rename autopilot to something like "lane assist" and consider a release of liability form" in plain English. Keep providing and improving what's there, but get out in front of this problem. I think you are legally in the clear, but
perception is becoming a problem, and people may not pay the 2500/3000 for the option. It works very well for me, under reasonable supervision.
 
My 2010 Prius had AEB that was designed exactly the same way. As far a i know, most AEB systems are designed this way, for good reason. The cost of a false positive (slamming on the brakes in heavy traffic) is extremely high from a liability perspective.
Is it really? Like I said, here in NY, if you get rear-ended, *it is the fault of the person who rear-ended you, period*.

By contrast, allowing the car to crash into the car ahead of you when you could have stopped it... exposes the company to extremely high liability.

I think they've assessed the liability risk backwards, bluntly.
 
Is it really? Like I said, here in NY, if you get rear-ended, *it is the fault of the person who rear-ended you, period*.

By contrast, allowing the car to crash into the car ahead of you when you could have stopped it... exposes the company to extremely high liability.

I think they've assessed the liability risk backwards, bluntly.
Actually I think you have it backwards. If an AEBS system had a false positive and braked the car suddenly such that the driver couldn't possibly have overridden, then the car company could possibly have liability since the system caused the accident.

On the flip side, if the AEBS failed to completely prevent an accident, the car company has no liability, since absent of the system, the driver would have crashed anyways. The AEBS system is only tasked with reducing the likelihood of an accident or damage/injury/death from an accident, it does not have to prevent all accidents.
 
Actually I think you have it backwards. If an AEBS system had a false positive and braked the car suddenly such that the driver couldn't possibly have overridden, then the car company could possibly have liability since the system caused the accident.

On the flip side, if the AEBS failed to completely prevent an accident, the car company has no liability, since absent of the system, the driver would have crashed anyways. The AEBS system is only tasked with reducing the likelihood of an accident or damage/injury/death from an accident, it does not have to prevent all accidents.
Yes, you expressed this much better than I did.
 
A good autopilot design would make sure airplanes would never crash and car would never collide.

That would be a perfect autopilot. That will not exist in our life times. Just like perfect seatbelts and air bags would prevent all injuries, and perfect tires would never go flat. That's why we still have pilots, doctors, an, in some cars, spare tires, for others, tow trucks.