Speak EV - Electric Car Forums banner

Disable autopilot and similar systems?

  • Yes, all off

    Votes: 4 7.3%
  • No, but need to improve counter measures quickly

    Votes: 11 20.0%
  • Temporary, until they can sort it out

    Votes: 8 14.5%
  • No, testing is necessary, casualties happen

    Votes: 32 58.2%

  • Total voters
    55
1 - 20 of 206 Posts

·
Registered
Joined
·
1,114 Posts
Discussion Starter #1
Reading the news today, I wanted to see what everyone else thinks about disabling autopilot in all teslas? As we have seen with uber, the autonomuouos systems are not up to the task yet. As we are all road users, I think that should be of concern to everyone on the road.

For the purists, I have added the option: all levels, all manufacturer's systems, all off.
 

·
Registered
Joined
·
13,806 Posts
Do you take all the cars if the road every time a human crashes one?

The NHTSA investigation is ongoing and no conclusion has been reached, so it’s a bit premature to start talking about banning things. In any case Autopilot has never claimed to be 100% fool proof and requires driver supervision. It tells you so everytime you engage it.

Tesla point out that there are hundreds of AP journeys per day on this exact stretch of road without incident. So where do you think the fault is likely to lie?
 

·
Registered
Joined
·
991 Posts
They need to stop calling it autopilot for a start. People are idiots as each one of these collisions demonstrates. Watching DVDs, trusting 'autopilot' on a section of road where the same drive had called out Tesla for issues with the autopilot on a previous occasion etc etc.
 

·
Registered
Joined
·
13,806 Posts
They need to stop calling it autopilot for a start. People are idiots as each one of these collisions demonstrates. Watching DVDs, trusting 'autopilot' on a section of road where the same drive had called out Tesla for issues with the autopilot on a previous occasion etc etc.
I’ve lost count of how many times we’ve gone over this, but AP in aircraft is little more than cruise control with altitude hold.

It won’t stop you hitting a mountain or another aircraft either.

I suppose they should stop calling it Cruise Control because too many people think it will brake for traffic lights and objects in the road.
 

·
Registered
Joined
·
7,609 Posts
What I don't understand is why AEB didn't kick in. The same happened with the car that rear-ended a fire truck.

Am I being too simplistic to think that if the forward facing radar senses any object with a differential velocity that can't be scrubbed off in the distance between itself and that object the car should put on the brakes?

Interestingly Waymo uses a second independent AEB system for exactly this reason.
 

·
Registered
Joined
·
7,609 Posts
Tesla point out that there are hundreds of AP journeys per day on this exact stretch of road without incident.
Yes but with what version of hardware and software, time of day, lighting conditions, traffic speed etc.

If your car successfully drove the exact same road flawlessly every day for weeks before, it would build up complacency. One OTA update later and the car behaves differently. Could this be a contributory factor?

We know the owner complained about AP a few days before, I wonder if it was updated to a later version?

I guess we will have to wait for the official report.
 

·
Registered
Joined
·
13,806 Posts
The attenuator may not have presented a very strong radar return and the AEB is probably tuned to recognise the return from another car.

The problem with radar systems is they all require some sort of filtering to exclude all the spurious returns. This could be from a cats eye or something which may be quite small but would provide a strong return signal, much like radar reflectors for yachts.

Also I see a lot of people trying to set off the AEB with plastic items, which is pretty much transparent to radar.
 

·
Registered
Joined
·
13,806 Posts
Yes but with what version of hardware and software, time of day, lighting conditions, traffic speed etc.

If your car successfully drove the exact same road flawlessly every day for weeks before, it would build up complacency. One OTA update later and the car behaves differently. Could this be a contributory factor?

We know the owner complained about AP a few days before, I wonder if it was updated to a later version?

I guess we will have to wait for the official report.
Again hearsay, but apparently the owner was complaining that his AP kept failing on this stretch of road.

So he should really have been aware of the issue. Looks similar to the other AP death where the driver was distracted for a prolonged period - apparently about 6 seconds.

We also don’t know what other traffic was doing at the time. For instance if the car in front swerved out of the way of the attenuator, then the systems and driver would have little time to react.
 

·
Moderator
Joined
·
9,719 Posts
Over 60mph AEB gets bad very quickly, any touch of the brake gives driver control and disables AEB.

Filtering static objects is a bloody hard problem, especially without LIDAR for a 3D map.
 

·
Registered
Joined
·
2,861 Posts
What I don't understand is why AEB didn't kick in. The same happened with the car that rear-ended a fire truck.
The problem with radar-based detection is the false-positives: it too easily triggers on stationary objects that aren't obstructions. Hence the standard approach of filtering the detections to remove stationary objects for ACC purposes and only respond using AEB at very close quarters where the object is so large and clearly in view that collision is inevitable.

Quite possibly in this case AEB did cut in but only in the last few metres where it would have negiligible effect at the speed in question. Quite possibly in the fire truck case it either did likewise, or the truck was only partly obstructing the lane and so never did appear a positive enough detection to trigger AEB. (purely speculation in these examples, I have no data).

Over 60mph AEB gets bad very quickly, any touch of the brake gives driver control and disables AEB.
Not according to the NTSB report of the Florida crash which explained that, in these cases where the radar has detected a potential obstruction but not triggered AEB due to ambiguity, driver's foot on the brake is considered to confirm the detection and trigger a 100% brake application.
 

·
Registered
Joined
·
769 Posts
The logic behind the question just seems completely bonkers...

Your suggesting disabling a system that has been extensively documented to massively decrease the risk of injury and death because of a single incident where it was active.

By the same logic you would also ban seatbelts, crash barriers, crumple zones and airbags because someone died in a car equipped with these features.
 

·
Registered
Joined
·
7,609 Posts
We also don’t know what other traffic was doing at the time. For instance if the car in front swerved out of the way of the attenuator, then the systems and driver would have little time to react.
The Tesla report claims the driver had a clear unobstructed view of the barrier for 5 seconds or 150m.

This suggests no cars swerved across the lanes. It also gives us a ballpark estimate that the car was travelling at between 65 and 70mph. (And also suggests that if the driver had been paying attention would have had sufficient stopping distance).
 

·
Registered
Joined
·
7,609 Posts
Your suggesting disabling a system that has been extensively documented to massively decrease the risk of injury and death because of a single incident where it was active.
Strictly speaking this is not entirely true.

What has been extensively proven is AEB has reduced rear end accidents by 40%. AFAIK there are no such studies for lane keeping technology yet. (Equipping vehicles with AEB helps reduce rear-end crashes by 40%)


Tesla latched on to this and stated AP equipped cars were 40% less likely to be involved in rear end accidents than earlier cars without AP hardware. I don't think anyone is advocating removing AEB.
 

·
Registered
Joined
·
13,806 Posts
Tesla latched on to this and stated AP equipped cars were 40% less likely to be involved in rear end accidents than earlier cars without AP hardware. I don't think anyone is advocating removing AEB.
Well for some reason insurance companies are offering discounts for AP equipped cars, but not those equipped with AEB. Why do you think that is? :)
 

·
Registered
Joined
·
249 Posts
The voting options sound like that trick question about "when did you stop beating your wife?"

someone died, so something obviously went wrong.
I don't think a "daily mail" style of poll is going to help the situation.

in this case, perhaps a better design of road layout might have helped?

there is rumor and speculation that there are regular crashes at this particular junction (from cars not on auto pilot), perhaps to the point where it is common for this crash barrier to be left unpaired for a number of days, perhaps because the repair crew are of the opinion that it is not worth wasting their time to keep repairing it, when it is only gong to get hit again?

a quick google of concrete blocks on motorway junctions led me to this
Motorway crash barriers - Brake the road safety charity

Since this incident happened in America, the land of "give guns to teachers to prevent school shootings", I think we need to wait a little longer for more (better?) information, in the hope that the real reason for this failure is revealed, rather than knee jerk reactions.
 

·
Well-Loved Member
Joined
·
985 Posts
Well for some reason insurance companies are offering discounts for AP equipped cars, but not those equipped with AEB. Why do you think that is? :)
Because without AP the crash stats are much worse than standard cars with AEB? Let's not forget what the IIHS said last year:

"Teslas get into a lot of crashes and are costly to repair afterward," said Russ Rader, spokesman for the Insurance Institute for Highway Safety, which is the Highway Loss Data Institute's parent organization. "Consumers will pay for that when they go to insure one."

Also when you claim "insurance companies are offering discounts" you mean one UK insurer as part of some ill-thought-out PR romance:

Direct Line reveals details of tie-up with Tesla

That AP discount doesn't sound quite so marvellous now, does it? And after this week's news it just looks... well, wrong. There is no proof that current autonomous tech makes the car safer.
 

·
Registered
Joined
·
13,806 Posts

·
Registered
Joined
·
778 Posts
Yes but with what version of hardware and software, time of day, lighting conditions, traffic speed etc.

If your car successfully drove the exact same road flawlessly every day for weeks before, it would build up complacency. One OTA update later and the car behaves differently. Could this be a contributory factor?

We know the owner complained about AP a few days before, I wonder if it was updated to a later version?

I guess we will have to wait for the official report.
I thought Tesla claimed he had complained about the mapping not AP? Whilst we have 2 accounts can we believe either 100%
 
1 - 20 of 206 Posts
Top