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

Data likely shows Teslas on Autopilot crash more than rivals

This site may earn commission on affiliate links.
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.
Without near real-time telemetry, large percentages of accidents will go unreported simply because the manufacturer will never know about them. I would venture that most automobiles are repaired by third party shops that have no reporting requirements even if there are suitable logs stored on the vehicle.

A car that crashes may not send telemetry if the crash damages equipment needed to send the data. So, some data may not be available unless the car is returned to a dealer or manufacturer for repair. This likely explains Tesla's not having the telematics column marked.
 

Level 2 ADAS Crashes by Reporting Entity​

Tesla

273
Honda

90
Subaru

10
Ford

5
Toyota

4
BMW

3
General Motors

2
APTIV

1
Hyundai

1
Lucid Motors

1
Porsche

1
Volkswagen

1
Source: National Highway Traffic Safety Administration

And what conclusion do you draw from these numbers?

Answer: None at all, since the relative populations of cars with ADAS per maker and miles driver are not taken into account.
 
And what conclusion do you draw from these numbers?

Answer: None at all, since the relative populations of cars with ADAS per maker and miles driver are not taken into account.
@2101Guy wants visitors to TMC to think that Tesla is not safe I.e. as usual a very dishonest post.

BTW, Waymo has the highest number of crashes among ADS - but he is not talking about that.
 
I’m almost 100% sure they’re not using results of FSD Beta. The production FSD build uses the same highway AP stack as basic AP, and frankly TACC uses the same accelerate/braking logic too. Their highway performance would all be identical.

FSD Beta is L2 so by the NHTSA instructions they have to report accidents that meet the reportable crash criteria.

It does add an interesting twist as it adds the possibility of accidents on city streets where a lot of accidents happen. Places where no other L2 car wouldn't likely have an ADAS system engaged.
 
  • Like
Reactions: sleepydoc
I’m almost 100% sure they’re not using results of FSD Beta. The production FSD build uses the same highway AP stack as basic AP, and frankly TACC uses the same accelerate/braking logic too. Their highway performance would all be identical.

Have you looked at the spreadsheet yet?

There is an alarming number of crashes reported with Tesla vehicles where the speed limit was 40 or less.

The Pre-Crash movements for some of them are a bit alarming, but the narrative is usually redacted so who knows what happened.

Line 106 is concerning because its on the street, and a Model S hit a pedestrian.
 
Last edited:
  • Love
Reactions: 2101Guy
It doesn't look like Honda used any telemetry, and instead relied heavily on customer reports.

From the customer reports its clear that in lot of cases the ADAS system had nothing to do with the accident, and in a lot of cases couldn't have even been activate due to the limitations of the Honda Sense system.

It's a good read if anyone has time to read through what customers have reported for Honda.
 
There’s an average of 17,000 reported car crashes every DAY in the United States. In 2020 there were 34,766 fatal car crashes for the year total (95/day) (Fatality Facts 2020: State by state).

Yet we have a government agency that wants to investigate ~400 crashes owing to the difference caused by FSD/AP. That’s 2% of the daily national car crash or 1% of the national yearly fatal crashes; even less if you factor in the fatal FSP/AP related crashes.

Progress is being impeded by the safety folks who fail to realize, or understand, the potential gains with FSD/AP. Humble thoughts here….
 
There’s an average of 17,000 reported car crashes every DAY in the United States. In 2020 there were 34,766 fatal car crashes for the year total (95/day) (Fatality Facts 2020: State by state).

Yet we have a government agency that wants to investigate ~400 crashes owing to the difference caused by FSD/AP. That’s 2% of the daily national car crash or 1% of the national yearly fatal crashes; even less if you factor in the fatal FSP/AP related crashes.

Progress is being impeded by the safety folks who fail to realize, or understand, the potential gains with FSD/AP. Humble thoughts here….
But NHTSA can't do anything about stupid, which causes the vast majority of crashes, so they investigate and regulate what (they at least think) they can control.

One of the primary differences about Tesla, and the source of MANY complaints, is that they label so many of their ADAS features "BETA," talk about their customers as "testers," and post warnings about how much the driver has to pay attention and how potentially dangerous theses BETA/untested ADAS features are. Given all that, I understand why they would come under unusual scrutiny from NHTSA. Is this the only way to be on the bleeding edge? Maybe. But it seems that we are beginning to see ADAS systems from Mercedes and GM that rival Tesla NoA but without all of these warnings or "testing" labeling.
 
Last edited:
At this point the only accurate information that can be taken from this report is that accidents have happened. The report itself says so. Since we really can’t say anything about rates, relative rates, frequencies, systems involved, etc. what’s the point of continuing the thread?

Because its comically bad.

I fully support what they're trying to do. To me its not about Tesla, but how humans use L2 technology.

The Tesla data shows how incomplete basic telematic data is without knowing the user story.
The Honda data shows how a user story doesn't line up at all with the capabilities of the system.
 
NHTSA has CSV files containing incident report data for both Level 3-5 ADS and Level 2 ADAS available for download: Standing General Order on Crash Reporting | NHTSA

Tesla seems to have provided both a 1-day and 10-day update for a vehicle crash under Level 3-5 ADS. It was a 2018 Model 3 RWD with the Consumer driver involved in an incident at April 2022 8:46pm during daylight in Coxs Creek, KY on a 55mph Rural Road on a Dry / clear weather with a Passenger Car ("crash partner") who was Passing and had contact area to its Front Left, Front, Front Right and airbags deployed while the Tesla ("subject vehicle") was Proceeding Straight at 28mph with contact area to its Left, Front Left, Front, Right, Front Right and airbags deployed.

Curious that this was filed under ADS as maybe that's suggesting FSD Beta? The crash seems to have been a head-on collision with the oncoming vehicle trying to pass or drifted out of lane with potentially AEB slowing down but not stopped.


Looking at the Level 2 ADAS report for Tesla, here's the listed Pre-Crash Movement of the "crash partner"
Code:
 374 Unknown
  77 <blank for Pole / Tree or Other Fixed Object>
  36 Proceeding Straight
  19 Stopped
  13 Making Left Turn
  12 Changing Lanes
   5 Lane / Road Departure
   4 Other, see Narrative
   3 Crossing into Opposing Lane
   2 Entering Traffic
   1 Parked
   1 Merging
   1 Making U-Turn
   1 Backing

And for the Tesla "subject vehicle"
Code:
 354 Unknown
 138 Proceeding Straight
  31 Lane / Road Departure
  10 Other, see Narrative
   5 Making Left Turn
   4 Changing Lanes
   3 Stopped
   2 Making Right Turn
   1 Traveling Wrong Way
   1 Backing

So even more curious that there wasn't even one "Passing" for Tesla ADAS reports although "Crossing into Opposing Lane" is probably similar but why not reuse that?
 
Last edited:
  • Informative
Reactions: EVNow and S4WRXTTCS
But NHTSA can't do anything about stupid, which causes the vast majority of crashes, so they investigate and regulate what (they at least think) they can control.

One of the primary differences about Tesla, and the source of MANY complaints, is that they label so many of their ADAS features "BETA," talk about their customers as "testers," and post warnings about how much the driver has to pay attention and how potentially dangerous theses BETA/untested ADAS features are. Given all that, I understand why they would come under unusual scrutiny from NHTSA. Is this the only way to be on the bleeding edge? Maybe. But it seems that we are beginning to see ADAS systems from Mercedes and GM that rival Tesla NoA but without all of these warnings or "testing" labeling.
If only there was a choice that the customer had to make…to elect to PARTICIPATE in BETA testing which is fully disclosed prior to their receipt of the FSD Software.

Few are asking what the drivers/owners are doing in these crashes. Were they doing their part, as they agreed to, in their choice To BETA FSD? I suspect not…and rather than hold them accountable, we will make life miserable for everyone else that is being responsible for their BETA role.

Otherwise, Tesla’s adaptive cruise control works just as well as Audi, MB, etc etc.
 
If only there was a choice that the customer had to make…to elect to PARTICIPATE in BETA testing which is fully disclosed prior to their receipt of the FSD Software.

Few are asking what the drivers/owners are doing in these crashes. Were they doing their part, as they agreed to, in their choice To BETA FSD? I suspect not…and rather than hold them accountable, we will make life miserable for everyone else that is being responsible for their BETA role.

Otherwise, Tesla’s adaptive cruise control works just as well as Audi, MB, etc etc.
Let's not conflate "Autosteer on City Streets," aka "FSD Beta," which is an actual beta test, and the features of "Full Self-Driving Capability," which is Telsa's name for their production ADAS features. You add the "Full Self-Driving Capability" to your car and get the ADAS features when it's delivered, just like the technology package of any other car.
 
But NHTSA can't do anything about stupid, which causes the vast majority of crashes, so they investigate and regulate what (they at least think) they can control.

One of the primary differences about Tesla, and the source of MANY complaints, is that they label so many of their ADAS features "BETA," talk about their customers as "testers," and post warnings about how much the driver has to pay attention and how potentially dangerous theses BETA/untested ADAS features are. Given all that, I understand why they would come under unusual scrutiny from NHTSA. Is this the only way to be on the bleeding edge? Maybe. But it seems that we are beginning to see ADAS systems from Mercedes and GM that rival Tesla NoA but without all of these warnings or "testing" labeling.
stupid, drunk, teenager, poor, elderly... can't regulate demographics
 
Interesting Volvo’s Pilot Assist isn’t mentioned. It’s basically Tesla’s AP1 system, but not as good. I had two Volvo’s with it, and it was pretty decent.

When the cars went into service, Volvo would send data back to the “mother ship”, and Volvo had updates they would apply when cars came in for service. Not OTA, but they would load the newest software onto cars if they were eligible at the time of service.

It was a long, drawn out process that required the car to be put onto a trickle charger while the software was being loaded. Mostly, it was stupid map updates that you would never know were even updated, but the invoice would say “software update”. A couple of times, the behavior of Pilot Assist would change, so they were doing something.

Apparently, now Pilot Assist reads map data and slows for curves. It’s called PA3 or something now. Not ground breaking, but promising in any event.
 
  • Informative
Reactions: EVNow