Blog Tesla Releases Data on Utah Autopilot Crash

Last week, a woman in Utah crashed her Autopilot-enabled Tesla Model S into the back of a parked fire truck at 60 mph. The car was totaled, but the woman escaped with only a broken ankle.

During an investigation of the crash, the woman admitted that she was looking at her phone during the accident. In addition to local law enforcement, the crash is also under investigation by the National Highway Traffic Safety Administration.

Tesla agreed to cooperate with investigators and on Wednesday, the South Jordan Police Department shared details from data recovered on the car’s computer.

Technicians from Tesla successfully recovered the data from the vehicle. According to Tesla’s

report, shared in a press release from the police department, the vehicle indicated:



The driver engaged Autosteer and Traffic Aware Cruise Control on multiple occasions

during this drive cycle. She repeatedly cancelled and then re-engaged these features, and

regularly adjusted the vehicle’s cruising speed.

Drivers are repeatedly advised Autopilot features do not make Tesla vehicles

“autonomous” and that the driver absolutely must remain vigilant with their eyes on the

road, hands on the wheel and they must be prepared to take any and all action necessary

to avoid hazards on the road.

The vehicle registered more than a dozen instances of her hands being off the steering

wheel in this drive cycle. On two such occasions, she had her hands off the wheel for

more than one minute each time and her hands came back on only after a visual alert

was provided. Each time she put her hands back on the wheel, she took them back off the

wheel after a few seconds.

About 1 minute and 22 seconds before the crash, she re-enabled Autosteer and Cruise

Control, and then, within two seconds, took her hands off the steering wheel again. She

did not touch the steering wheel for the next 80 seconds until the crash happened; this is

consistent with her admission that she was looking at her phone at the time.

The vehicle was traveling at about 60 mph when the crash happened. This is the speed

the driver selected.

The driver manually pressed the vehicle brake pedal fractions of a second prior to the

crash.

Contrary to the proper use of Autopilot, the driver did not pay attention to the road at all

times, did not keep her hands on the steering wheel, and she used it on a street with no

center median and with stoplight controlled intersections.



Police said the driver of the Tesla was issued a traffic citation for failure to keep proper lookout under South Jordan City municipal code 10.28.030 (traffic infraction).

“As a reminder for drivers of semi-autonomous vehicles, it is the driver’s responsibility to stay

alert, drive safely, and be in control of the vehicle at all times,” the release said. “Tesla makes it clear that drivers should always watch the road in front of them and be prepared to take corrective actions. Failure to do so can result in serious injury or death.”

NHTSA continues to conduct their own review of this incident.

 

[ThosEM]

Read the manuals of other systems. You see the same issue. From the 2018 Infiniti Q70 owner’s manual:

The radar sensor will not detect the follow- ing objects:
. Stationary and slow moving vehicles

The Doppler shift for the road is the same as for the stopped vehicle. It is a problem of missed positives vs. false positives. Imagine if every time you approached a hill, the car started to brake automatically.

I've read about that and I get it, kind of. Still, it seems to me that a vehicle (or any other object) standing in the road immediately ahead should have a different and easily discriminated sensor signature than a simple change of slope of the road. Until it does, I don't think driverless cars will be going anywhere useful.

 

[benfar]

I've read about that and I get it, kind of. Still, it seems to me that a vehicle (or any other object) standing in the road immediately ahead should have a different and easily discriminated sensor signature than a simple change of slope of the road. Until it does, I don't think driverless cars will be going anywhere useful.

I'm not familiar with the Infiniti's radar, but the Bosch MRR in Hardware 1 Teslas and the Continental ARS410 in Hardware 2 and 2.5 Teslas are more sophisticated than a simple Doppler radar. Plus, starting with software 8.0 Tesla added technology to improve stationary-object detection for their cars with either of those radars ( Upgrading Autopilot: Seeing the World in Radar ).

The recent Insurance Institute for Highway Safety tests of the (Hardware 2.5) Model 3's automatic emergency braking for a stopped vehicle ( IIHS rates Tesla Model 3 automatic emergency braking and forward collision warning ) showed it completely avoided a collision from an initial speed up to 25 mph, the highest initial speed the IIHS currently tests.

 

[croman]

Just a slight correction, only AP2.5 uses the Continental radar. AP2 uses the Bosch MRR.

I've found untracked vehicle detection and action to be nearly flawless up until 50mph. After that and its dicier than I am willing to test. Occasionally between 40-50mph on turns it will not react soon enough for me but that's very rare after 2018.10.4 (currently on 2018.18 (AP2 HW)).

 

[benfar]

Just a slight correction, only AP2.5 uses the Continental radar. AP2 uses the Bosch MRR.

I've found untracked vehicle detection and action to be nearly flawless up until 50mph. After that and its dicier than I am willing to test. Occasionally between 40-50mph on turns it will not react soon enough for me but that's very rare after 2018.10.4 (currently on 2018.18 (AP2 HW)).

You're absolutely right about the Hardware 2 radar. I knew better than I typed.

 

[ThosEM]

Just a slight correction, only AP2.5 uses the Continental radar. AP2 uses the Bosch MRR.

I've found untracked vehicle detection and action to be nearly flawless up until 50mph. After that and its dicier than I am willing to test. Occasionally between 40-50mph on turns it will not react soon enough for me but that's very rare after 2018.10.4 (currently on 2018.18 (AP2 HW)).

The only AP experience I have is with a couple of loaner vehicles, but driving them on the busy multi-lane DC beltway made me acutely aware that AP is a very long way from "self-driving". It isn't sufficient for the forward awareness to be limited to the next one or two cars or ahead. A human driver on a busy highway routinely sees all the brake lights ahead, for hundreds of meters, turning red whenever there is an issue that is slowing or stopping traffic. AP has no awareness of that. It's really just traffic-aware cruise control fused with automated steering between the lines.

I suppose that translates directly into a speed limit on the effectiveness of emergency braking. AP can't avoid what it can't see in time to take effective action. The faster you go, the longer it takes to stop, and AP doesn't really see more than a few cars ahead. I have a hard time imagining that it could swerve to avoid something for which it doesn't have time to stop.

Autopilot in an aircraft has no requirement to deal with a sudden requirement to slow or stop or swerve. Similarly, autopilot in a Tesla has no ability to deal with a sudden requirement to slow or stop or swerve. That tells me that AP might be great for an airplane but it has no business on a highway without close supervision (as Tesla specifies!), with the possible exception of navigating a slow bumper to bumper traffic jam.

All the that is to some degree just confirmation bias in that I have not so far paid for AP, and probably will not, in the foreseeable future.

 

[bro1999]

Tesla in Autopilot mode sped up before crashing into stopped fire truck, police report says

and
The driver of the vehicle, Heather Lommatzsch, 29, told police she thought the vehicle's automatic emergency braking system would detect traffic and stop before the car hit another vehicle.

 

[seattlite2004]

Tesla in Autopilot mode sped up before crashing into stopped fire truck, police report says

and
The driver of the vehicle, Heather Lommatzsch, 29, told police she thought the vehicle's automatic emergency braking system would detect traffic and stop before the car hit another vehicle.

Sounds like user error to me. With regular cruise or even ACC, if the car speeds up into wall/object....its the driver's fault....why should AP be any different. Driver was an idiot...hope she learned a lesson.

 

[SO16]

Are there any AEB systems that would stop from 60 MPH to zero?

 

[Electroman]

Are there any AEB systems that would stop from 60 MPH to zero?

Nope. Not for a stopped object, that I am aware of in any car.

Moving object sure. That happens everyday in Tesla and many cars that have TACC.

 

[SO16]

Nope. Not for a stopped object, that I am aware of in any car.

Moving object sure. That happens everyday in Tesla and many cars that have TACC.

I didn’t think so either. I couldn’t find any.

 

[Electroman]

So in terms of TACC capability it is on par or have a slight edge over other cars. Autosteering is of course unmatched.

We have not heard of L2 level accidents in MBZ, BMW, Nissan and Chevy because very few use these functions in general as they are subpar.

 

[EVie'sDad]

Does anyone think that if the driver (insert latest accident victim's name here) had been paying attention to the road and what was in front of them, that things might have turned out differently?

That we would not be reading endless articles about negligence, irresponsible behavior and tragedy?

I believe for now, we all need to agree that driving with AP1 or AP2 is more challenging and requires constant monitoring.

Vigilance is mandatory as the auto steer and TACC will not respect red lights, Gore zones, stopped vehicles or other large obstructions or construction areas. Nothing wrong with that, as it is considered only level2 driving assisting.

The problem is with drivers expectations, ignore this at your own peril. We cannot fix stupid or irresponsible behavior, nor can any car manufacturer.

If you don't agree with this, then don't use it.

 

[SR22pilot]

I've read about that and I get it, kind of. Still, it seems to me that a vehicle (or any other object) standing in the road immediately ahead should have a different and easily discriminated sensor signature than a simple change of slope of the road. Until it does, I don't think driverless cars will be going anywhere useful.

If I understand the scenario in question correctly, it is a very specific one where the radar is properly tracking a moving vehicle. The moving vehicle moves to the side and uncovers a stopped vehicle. That will result in an increase in the signal from the road if there isn’t a stopped vehicle. The stopped vehicle should generate an even bigger increase in signal but one matching the road in Doppler shift. There is probably a solution that will get implemented down the road but it is key to avoid false positives since that is also dangerous. It gets more complex when you realize you don’t want to brake for signs on the side of the road, overhead signs or bridges.

 

[SR22pilot]

Does anyone think that if the driver (insert latest accident victim's name here) had been paying attention to the road and what was in front of them, that things might have turned out differently?

That we would not be reading endless articles about negligence, irresponsible behavior and tragedy?

I believe for now, we all need to agree that driving with AP1 or AP2 is more challenging and requires constant monitoring.

Vigilance is mandatory as the auto steer and TACC will not respect red lights, Gore zones, stopped vehicles or other large obstructions or construction areas. Nothing wrong with that, as it is considered only level2 driving assisting.

The problem is with drivers expectations, ignore this at your own peril. We cannot fix stupid or irresponsible behavior, nor can any car manufacturer.

If you don't agree with this, then don't use it.

I agree. How about the driver who had complained something like 7 times that his car performed improperly and dangerously at a spot on the road and then gets killed at that spot when AP doesn’t work correctly. He KNEW there was an issue there and yet wasn’t paying attention.

 

[Reciprocity]

Here is the question that I have and I think I know the answer. In general the AP is safer because when AP is engaged the car will not become distracted, where as the normal driver is distracted dozens of times per hours, maybe even more. Those distractions very of course from very minor to a few seconds looking at a phone. AP is safer because these 100 distractions are covered nearly flawlessly. The problem is that people get a heightened sense of security with AP and become a bit more reckless. I can admit to this myself and I have to remind myself to be more diligent all the time. I tend to use AP very often and I like to scan the road farther ahead, rather then focus on the lines and car directly in front of me. I think this makes it safer as well.

The question I have is this.. Is AP safer then no AP for the average driver? If Tesla made it more annoying with nags would it be used more or less to the betterment or determent of the owner?

In theory the Nags could learn your pattern and only mess with you if you are clearly abusive. I can see in this case specifically and the in the case in mountain view, Tesla has detailed data on how many times the person had to be nagged and how quickly they removed their hands from the steering wheel. Play fair, and less nags. Be a giant a-hole and get many more nags and even lock outs where you have to pull over and park to use AP again.

Overall I think AP is much safer, but its not perfect and wont be for close for a while. NO system will ever be PERFECT but AP with FSD will be close enough to save many lives. Remember, even if you dont have FSD, it will still be there watching and able to take emergency actions. For example, you are about to run a red light, or blow through a stop sign with a cyclist approaching.

 

[Electroman]

I believe for now, we all need to agree that driving with AP1 or AP2 is more challenging and requires constant monitoring.

Not True. Not at all true. I specifically take exception to the 'more challenging' characterization. Not going to let these kind of false statements go unchallenged.

AP requires much less attention than normal driving. With my extensive driving experience with AP, i would make an attempt of defining a complexity number system something like this. Think of this as Gartner Quadrant.

Scale:
10 - for driving a car with no AP
0 - for being a passenger and not driving.

Different scenarios:
3 - Highways stop and go traffic (0 to 20 mph) : you can lazily watch the traffic and even take eyes of the road for a few moments. The only concern is when someone butts into your lane and only one wheel inside your lane.

5 - Highways, 60 mph+ with clear lines and no confusion around exits or lane splits or mergers - essentially center lane driving. There is really nothing much to do here except watch for any dead stopped traffic. This is 90% of my commute every day. You have to watch for construction zones that might come out of nowhere overnight.

7 - Highways 50 mph+ with tangential exits, splits and mergers. You need to watch and be ready to take over of needed. A little bit of commonsense and experience with AP will tell you which are trouble spots for AP.

8 - Familar backroads less than 50mph with clearly marked lines. Watch for cutting traffic, cross traffic, stop lights and stopped vehicles and sharp curves. If you are familiar with that road you would enjoy AP.

11 - Unfamilar backroads. You never know where the next light is, or where the next lane drop is or where the left turn lane comes out of nowhere. Why bother? just drive yourself.

11 - poorly marked lines to no lines. Requires more attention than driving without AP. Not worth the trouble.

13 - construction zones. Don't even bother. Just drive.

As you can see AP is a great stress reliever and very enjoyable in a variety of daily situations. Just pay attention when it is needed. Takes a bit of experience and a bit of COMMON SENSE.

A typical drive would vary between 3 and 13. I don't let AP drive in places where the attention requirement gets to 8 or more. For each person depending on their experience with AP, their number might vary. You start with 13 (or more) when you are new to AP. Nervous wreck. But gradually get better.

Tesla's goal is to get these numbers down for every scenario, one step at a time.

 

[McRat]

Are there any AEB systems that would stop from 60 MPH to zero?

Yes, cars with ACC stop from 60mph when the adaptive cruise is set. At least mine do.
Without the ACC hot, the radar is still painting out well past a football field . It will grab your attention with hellfury if you do not slow down quick enough for stopped traffic. I don't know or care if it hits the brakes at that stage. I'm not going to ignore the 'not exactly subtle' warning to hit the brakes just to answer internet questions. Because it won't be believed it does stop in time, and if doesn't stop in time, I wreck. Sort of a no-win situation.

That's the truth. You'll hear other things about the limitations of automotive radar and cameras on this site. Mostly because they forget that it's an dual tech integration for the AEB/FAB, both camera and radar are used to resolve what is a threat. There will always be exceptions to the rule, but many are bogus. Like signs and bridge supports mistaken for cars.

One specifically that needs to addressed is the "hidden stopped car" threat.

• You are driving down the highway with a car in front of you.
• There is a broken down/stopped car in the lane.
• The car in front of you moves, so you ram the stopped car.
• This is said to be a limit of technology, not something else.

Now wait a minute. This has been a possible scenario since multilane highways were first built, and was easily defeated. Why is it an issue now?

It's not. You were following too close, and you were not scanning with your eyes. You were told this years ago. The 2 second rule didn't fly out of monkey's arse. It allows for such a situation. And it will work with digital cameras and radar as well. AEB is not for tailgating. Tailgating always has a frontal impact threat even without a hidden car. You get behind a car that halts in 90ft from sixty (give a C7 some room) and your reaction time won't cover you.

 

[jgs]

I believe for now, we all need to agree that driving with AP1 or AP2 is more challenging

I couldn’t disagree more.

and requires constant monitoring.

Now, that part I do agree with.

 

[EVie'sDad]

I couldn’t disagree more.

Now, that part I do agree with.

Whether or not you agree is irrelevant, if you pay less attention, you do so at your own risk. The car isn't going to save you from yourself. It is only level 2 assistance, thus the driver is supposed to take control in any circumstance the vehicle cannot handle. Right now we must pay closer attention when such driver assistance devices are enabled, otherwise we may not live to see levels 3, 4 or 5.

What’s the Difference Between Level 2, Level 3, Level 4, and Level 5 Autonomy?

Hey, here's an idea, the next time you are in a normal Ice vehicle, try enabling cruise control then look away for a minute or two on the highway. See how many times you can do that. If you ignore it long enough, it too can kill you. Auto steering and TACC are in the same class. More vigilance, not less is requested for now until the technology becomes demonstrably proven to be safer when ignored.

 

[life_sav_er]

This is insufficient explanation, they need to explain *why* is happening....is it a sensor problem? Software problem?...how come the radar is not able to detect a stationary object like a firetruck....what are the implications for Full Self Driving.

Looking at all the accidents with Teslas in Autopilot mode, I'm quite sure I can explain the technical reason why we see these accidents. It's a combination of poor capabilities of the radar sensor (which is completely different than our visual perception) plus the design decision of Tesla to only brake (while driving at more than 50 mph) when both sensors (camera and radar) "see" the object. I hope this information reaches the Tesla community and can help to save lives.

"Old" radar sensors do not see stationary objects at all because there are so many reflections coming from everything around the vehicle that is in motion that they need to be filtered out (because there are so many).

Tesla has now upgraded the autopilot - see: Upgrading Autopilot: Seeing the World in Radar
But the Autopilot still relies on the (upgraded) radar sensor for braking decisions so it uses a sense we humans just do not have practical comprehension for. We see a flat metal surface with yellow and black stripes on it and we know that is a hard obstacle we do not want to run into - like a lane divider. The orientation of the surface is hardly of any relevance for the human eye. For radar the orientation is essential. If it is not perpendicular to the line of approach it is a "radar mirror" reflecting the radar beam away from the sensor. As a result the obstacle itself (and also everything behind it) is not detectable by the radar sensor. I'm convinced that that's the key contributor to the fatal accidents that killed Joshua Brown and Wei Huang. Flat non-perpendicular metal surfaces are invisible for the radar sensor. I think every Autopilot user should really know about this - and rethink where to put his or her attention while driving with Autopilot engaged.

You can see some pictures of the real situations which led to accidents here:
A1 Online-Festplatte

Cheers,
and trying to save lives