Data likely shows Teslas on Autopilot crash more than rivals

[EVNow]

Wait - are you telling me @2101Guy made a post with a dramatic, click bait headline about something reflecting poorly on Tesla that lacked context or nuance and omitted pertinent details? I am shocked. SHOCKED, I tell you!

@2101Guy apparently didn't read this important thing. OP won't be winning any honesty rewards any time soon.

This new data is the first of its kind, and the reports detail several important caveats and limitations to this dataset for researchers, the press and the public to consider. For a clear understanding of the data, users should read about the data limitations and the sources that manufacturers and operators used to collect and report crashes.

Further, these data are not normalized by the number of vehicles a manufacturer or developer has deployed or by vehicle miles traveled. That information is held by manufacturers and not currently reported to NHTSA. Thus, these data cannot be used to compare the safety of manufacturers against one another.

 

[Dewg]

Some people just want to watch the world burn... Half empty, Half full - screw it, throw the glass and piss on the wreckage.

 

[EVNow]

I haven't seen the stats for airbag deployments in the general auto fleet that would be needed to compare to Tesla's 4 million mile number. And of course you would also have to correct for driver demographics, location, time of day, etc. to get a meaningful comparison.

The fact is - the kind of data that is needed for actual comparison is simply not there. Forget "driver demographics, location, time of day, etc. " .... the biggest issue is all crashes are simply not reported. Most police departments will not even register a crash unless certain minimum $ damages or injuries are reported. Even insurance companies will not hear about any small crash for which the person doesn't file a claim. Only the Virginia Tech study that I've cited before has detailed enough information on all crashes.

With tesla, they do not track small crashes like hitting a curb. Yet, Tesla probably has the most detailed telemetry of all OEMs. I assume this statement refers to Tesla.

For example, some reporting entities provide the agency with robust data more quickly because their vehicles are equipped with telematics capabilities.

Manufacturers of Level 2 ADAS-equipped vehicles with limited data recording and telemetry
capabilities may only receive consumer reports of driving automation system involvement in a
crash outcome, and there may be a time delay before the manufacturer is notified, if the
manufacturer is notified at all.

Due to variation in data recording and telemetry capabilities, the Summary Incident Report Data
should not be assumed to be statistically representative of all crashes.

For example, a Level 2 ADAS-equipped vehicle manufacturer with access to advanced data
recording and telemetry may report a higher number of crashes than a manufacturer with limited
access, simply due to the latter’s reliance on conventional crash reporting processes. In other
words, it is feasible that some Level 2 ADAS-equipped vehicle crashes are not included in the
Summary Incident Report Data because the reporting entity was not aware of them.

 

[bfayer62]

Toyota, 4 crashes on adaptive cruise control? Who would have known, I know the owners of two of them.....

 

[EVNow]

My guess - only Tesla, Honda and Subaru have automated telematics to collect data.

Does anyone know where Honda & Subaru have automatic telematics ?

 

[EVNow]

Oh - among L3+ vehicles - obviously Waymo is the most dangerous

BTW, who is Transdev and why do they crash so much ?

 

[sleepydoc]

My guess - only Tesla, Honda and Subaru have automated telematics to collect data.

Does anyone know where Honda & Subaru have automatic telematics ?


View attachment 816981

We have a 2020 Subaru Forester. It has adaptive cruise and lane keep assist and an optional ‘star link’ system that requires a subscription. I have no idea if Starlink reports such data but I do know it’s not required so any reporting will be incomplete at best.

 

[bfayer62]

My guess - only Tesla, Honda and Subaru have automated telematics to collect data.

Does anyone know where Honda & Subaru have automatic telematics ?


View attachment 816981

I know the "STARLINK" system in my Subaru phones home, because they send me status reports on my Ascent. Check engine light, low tire pressure, etc. Once would assume they are collecting a lot of data on the backend.

 

[n.one.one]

Interesting video on the subject by Dr. Know-it-all Knows it all (38.1K subscribers).

 

[Dewg]

We have a 2020 Subaru Forester. It has adaptive cruise and lane keep assist and an optional ‘star link’ system that requires a subscription. I have no idea if Starlink reports such data but I do know it’s not required so any reporting will be incomplete at best.

Do you mean OnStar? I thought Starlink is the satellite internet service.

 

[TunaBug]

My guess - only Tesla, Honda and Subaru have automated telematics to collect data.

Does anyone know where Honda & Subaru have automatic telematics ?


View attachment 816981

My wife has a 2020 Honda Insight. I had no idea that it phoned home, but agree with your assumption based on the graph.

I drove that car almost exclusively for 4 months after she bought it (and before I bought my Model 3). The cruise control is much smoother than Tesla's TACC, although I don't like the UI as much. The lane keeping system is also smoother, but it wont engage below 45mph and it is not capable of turning the wheel very far: a few degrees left or right to follow most of the gentle highway curves, but when the highway makes an actual turn then you have to take over. Or to summarize: it is less capable than my Tesla, but it's nicer at the things it does do.

She _never_ uses either the cruise control or the lane keeping automation, unless by accident, and then she gets kind of panicked about "WTF IS IT DOING!?!?!?".

 

[dgatwood]

"Nissan, with over 560,000 vehicles on the road using its ”ProPilot Assist,” didn’t have to report any crashes, the company said."

I find this highly suspect especially if adaptive cruise control accidents count as well.

Not sure about the current version, but Nissan's ADAS from a few years back was just awful. Having driven with it, it is in many cases worse than not having any assist at all, wandering back and forth in the lane, turning off suddenly and without warning as you approach random intersections and off-ramps, (resulting in unexpected steering changes), etc. I'm assuming newer versions are better, but remember that unlike Tesla, those systems aren't getting over-the-air firmware updates, so the older ones are out there on the roads being actively dangerous, and will be until the last one of them gets sold for scrap metal. I have a very hard time believing that those have fewer crashes than AutoPilot... unless they aren't being reported because of a loophole.

Not giving the NHTSA nearly enough credit, these are people who specifically work in Auto safety and who surely know more about the relevant metrics than us observers who don’t work full time jobs in this very specialized sector

These are people who likely came from major automakers, as part of a revolving door between industry and regulation. That means they're motivated to find problems with Tesla, and a lot of them likely hold a grudge against Tesla for leaving their prior employers in the dust.

Cumulative miles travelled with the system
engaged is one of many figures requested last year by the NHTSA, you can find an exhaustive list by reading through their data request template

Cumulative miles driven with the system active is inherently misleading unless you're comparing very similar systems. The other ADAS systems I've used so far are limited to specific roads. Tesla's system, although being not intended for use on other roads, will provide lane keeping on city streets and minor highways, usually without much trouble, so long as it has at least something resembling reflector dots or lane lines on the left side. Thus, comparing crash counts per mile will lead to wildly misleading statistics that make Tesla appear worse than their competitors, while actually just being an indicator of how radically different the usage is. Those differences may or may not be reflected in covered miles.

The only fair comparison is crashes per million miles driven, without regard to whether the ADAS system is active. That approach eliminates any bias caused by some systems being able to ease the burden of city-street driving, where accidents are disproportionately high, and other systems refusing to do so. It also means that if driving with automation makes you less capable of driving safely when automation is unavailable, that will negatively impact the ADAS's score.

 

[TunaBug]

Oh - among L3+ vehicles - obviously Waymo is the most dangerous

BTW, who is Transdev and why do they crash so much ?

View attachment 816983

What got my attention in that graph is Apple's presence.

 

[sleepydoc]

Do you mean OnStar? I thought Starlink is the satellite internet service.

Subaru STARLINK In-Vehicle Technology

Subaru STARLINK connects drivers with safety and security services, touchscreen navigation, Bluetooth and smartphone applications like Pandora and iHeartRadio.

www.subaru.com

 

[sleepydoc]

What got my attention in that graph is Apple's presence.

Me too. Since they don't sell any cars the denominator is zero so....

 

[dgatwood]

Me too. Since they don't sell any cars the denominator is zero so....

That should make their vehicles infinitely deadly, no?

 

[EVNow]

Me too. Since they don't sell any cars the denominator is zero so....

That should make their vehicles infinitely deadly, no?

This is the ADS report - not ADAS i.e. these are L3+ vehicles a.k.a. self-driving cars.

Basically, mostly various companies testing their AVs on public roads. We know Apple has been testing their AV on public roads for some years now (because of CA reporting).

 

[EVNow]

Do you mean OnStar? I thought Starlink is the satellite internet service.

OnStar is GMs.

Subaru's is indeed called Starlink (the same name as SpaceX internet service).

Subaru STARLINK In-Vehicle Technology

Subaru STARLINK connects drivers with safety and security services, touchscreen navigation, Bluetooth and smartphone applications like Pandora and iHeartRadio.

www.subaru.com

 

[Supcom]

Is this correct? From my reading of this the NHTSA requires all significant accidents to be reported where any part of the L2 system was engaged.

This makes sense as an alert driver will prevent an accident from occurring regardless of who or what caused the situation from happening.

Phantom braking is not, per se, and accident. If PB results in an accident, then it would certainly be reportable.

 

[S4WRXTTCS]

My guess - only Tesla, Honda and Subaru have automated telematics to collect data.

Does anyone know where Honda & Subaru have automatic telematics ?

Have you looked at the raw data?

Column CY shows if the data came from telematic.

Sorting by Reporting Entity reveals that Honda doesn't have any reports via telematics.

The reason they have 90 reports seems to largely be related to the fact that there are around 5 million Honda vehicles with Honda Sense on the roads. So that 90 is extremely undercounted.

Subaru seems to have some telematics as they do have reports with CY as a yes.

A lot of Tesla's don't have telematics which is interesting.