Fatal autopilot crash, NHTSA investigating...

[Chickenlittle]

The Board doesn't seems to like Tesla Steering Wheel Sensor method:

"9. The way that the Tesla Autopilot system monitored and responded to the driver’s interaction with the steering wheel was not an effective method of ensuring driver engagement."

And its rationale is:

"6. Because driving is an inherently visual task and a driver may touch the steering wheel without visually assessing the roadway, traffic conditions, or vehicle control system performance, monitoring steering wheel torque provides a poor surrogate means of determining the automated vehicle driver’s degree of engagement with the driving task."

Does that mean the Board prefer GM Cruise eye-tracking monitoring system?

What most people forget but every high school student has mastered is that it is possible to stare at teacher and be elsewhere in thought. Eyes on road just as hands on wheel is no guarantee to prevent or prove attentiveness. People responsible for behaving responsibly and should be responsible if not. Car manufacturers are not responsible for customers inatentiveness

 

[Tam]

...Speling Polease...

Thanks for the clarification.

I mean 2018 GM "Cadillac Super Cruise™" with a camera on top of the steering column to track driver's eyes:

 

[stopcrazypp]

Sorry wrong thread.

 

[Tam]

...Eyes on road just as hands on wheel is no guarantee to prevent or prove attentiveness...

Very good point. I myself prefer to focus on the accident avoidance competency of the system rather than diverting effort to a nanny system to prove that hands-on-wheel and eyes-on-road is a proof of driver competency.

 

[bro1999]

AP 2.5 already has the interior cam so that is taken care of.

Why would someone dislike this post? Lol

 

[Economite]

Very good point. I myself prefer to focus on the accident avoidance competency of the system rather than diverting effort to a nanny system to prove that hands-on-wheel and eyes-on-road is a proof of driver competency.

Lane keeping + TACC ("autopilot") doesn't really have any accident avoidance compentency. AEB does avoid (or at least mitigate) accidents, but that feature is a passive safety feature that is always on, whether the driver turns on AP or not. Lane Keeping + TACC is basically a convenience feature.

 

[MP3Mike]

Why would someone dislike this post? Lol

Maybe because it doesn't. The Model 3 does, but AP2.5 in the S/X do not have an interior camera. (At least as far as what we have seen.)

 

[chillaban]

Lane keeping + TACC ("autopilot") doesn't really have any accident avoidance compentency. AEB does avoid (or at least mitigate) accidents, but that feature is a passive safety feature that is always on, whether the driver turns on AP or not. Lane Keeping + TACC is basically a convenience feature.

That doesn't make any sense. Having the car automatically regulate distance and steering plays a pretty key role in preventing fatigued/distracted drivers from triggering AEB/FCW scenarios anyway. Maybe in some cars where ACC is restricted to 30% of braking force and only looks 200ft ahead, but Tesla TACC is quite proactive and long-distance, and easily plays a safety role too.

 

[Economite]

That doesn't make any sense. Having the car automatically regulate distance and steering plays a pretty key role in preventing fatigued/distracted drivers from triggering AEB/FCW scenarios anyway. Maybe in some cars where ACC is restricted to 30% of braking force and only looks 200ft ahead, but Tesla TACC is quite proactive and long-distance, and easily plays a safety role too.

No question that TACC has value. Not sure that lane keeping has any safety value. If you're dozing off, you shouldn't be driving. If anything TACC + lanekeeping actually encourages folks to keep driving when they should be pulling off the road.

 

[malcolm]

Mind off you are wrong, that one is full autonomy so it doesn't require driver alertness.

I don't think it'll be so clear cut.

What independent testing will be in place to confirm a manufacturer's claims of Level 5?

Will it be a similar situation that has developed with emissions testing, where car makers simply create systems which can pass the required "Level 5 test"?

Full autonomy is a tough challenge. Sufficient autonomy is always going to be easier.

 

[Tam]

Lane keeping + TACC ("autopilot") doesn't really have any accident avoidance compentency...

Just because Autopilot is classified as "convenience", there is no reason that its function can not be improved to avoid accidents.

It has done that for me already: Such as it has applied braking on its own to slow down to a halt from full freeway speed at a stop-and-go traffic jam without my intervention at all.

Now, the effort needs to focus on how it can do it better and reliably even when there's another Florida scenario as discussed in this thread.

 

[Economite]

Just because Autopilot is classified as "convenience", there is no reason that its function can not be improved to avoid accidents.

It has done that for me already: Such as it has applied braking on its own to slow down to a halt from full freeway speed at a stop-and-go traffic jam without my intervention at all.

Now, the effort needs to focus on how it can do it better and reliably even when there's another Florida scenario as discussed in this thread.

Once again, this is AEB and/or TACC. Adding lane keeping to the mix isn't really a safety feature. Hard to come up with a safety benefit (let alone a significant safety benefit) from adding lane keeping.

 

[Tam]

...AEB...

Tesla AEB is not designed to brake to a halt.

When my Tesla applies brake on its own without my intervention to a complete stop from 70 MPH in order to avoid slamming into a freeway traffic jam in front, this is clearly not a function of Tesla AEB as stated in owner's manual:

...Hard to come up with a safety benefit (let alone a significant safety benefit) from adding lane keeping.

I think what you are trying to say is: Autosteer is not a safety feature.

It may not be for now, but someday, it will be able to steer into a freeway exit correctly so that human don't have to take risk to swerve into an exit at the last second due to inattentiveness.

Even with this current function, I think Autosteer is doing a better job than human is as I notice quite a few instances when other cars keep intruding lane markers, some white lines, some yellow lines.

That, to me, is a safety issue that Autosteer is addressing.

 

[chillaban]

Not departing your lane (e.g. properly functioning lane keeping) is absolutely a safety feature. There are many driving situations (e.g. around a cliff, hilly shoulders, stopped car on shoulder) where departing your lane can be a accident-causing, unrecoverable event. A car with the ability to automatically hold your lane with human or better-than-human abilities is absolutely a safety feature.

Note that with that said, I don't think Autosteer in AP1 or AP2 is there yet. I'm talking about the hypothetical world where there truly exists a lane holding system that's as good as a human.

 

[Economite]

If it's not there yet, then by my book it is not fair to currently call it a safety feature.

 

[WannabeOwner]

If you're dozing off, you shouldn't be driving.

Yes, but IME (and I didn't believe it until I had experienced it myself on multiple occasions). AP + long Journey definitely reduces my fatigue, and therefore means I finish my journey safely whereas before, regular journey, same time of night, quite late, before-AP I frequently fought sleep for the last 10 miles.

If anything TACC + lanekeeping actually encourages folks to keep driving when they should be pulling off the road.

if I was to be in a position where I ought to stop what would I do? If close to home I would probably "carry on". No difference (in making that decision) whether I have AP or not. But with AP at least I'm less at risk.

If not close to home, and therefore I must stop, what then? With AP it would beep at me when I nodded off, and wake me up. How many times would that have to happen before I decided I was not fit to drive? One would probably do me, perhaps a couple before I was convinced, so in that situation I would remain safe until I pulled over, and rest, rather than have an accident.

 

[Chickenlittle]

What most people forget but every high school student has mastered is that it is possible to stare at teacher and be elsewhere in thought. Eyes on road just as hands on wheel is no guarantee to prevent or prove attentiveness. People responsible for behaving responsibly and should be responsible if not. Car manufacturers are not responsible for customers inatentiveness

I am perplexed on how car companies make sure your not distracted in cars withOUT AP systems. Seems a lot larger problem (based on numbers of cars and lack of car taking notice). Why don't they demand automakers put in systems to avoid distracted drivers on ALL cars?

 

[AnxietyRanger]

Yes, but IME (and I didn't believe it until I had experienced it myself on multiple occasions). AP + long Journey definitely reduces my fatigue, and therefore means I finish my journey safely whereas before, regular journey, same time of night, quite late, before-AP I frequently fought sleep for the last 10 miles.

Had this been about AP1, I would agree. Generalizing it to include AP2, though, is not IMO correct. It is a very tiresome and stressful companion much of the time.

 

[Kant.Ing]

That's why in AP2.5 a driver-facing camera is added to bring us the telescreen from Oceania!

 

[WannabeOwner]

It [AP2] is a very tiresome and stressful companion much of the time

Sorry to hear that, I have no experience of AP2 but here's hoping that it soon levels out with AP1, and then surpasses it, very soon.