40% accident reduction since Autopilot's introduction - this will change landscape

[calisnow]

I think it's hard to overestimate the political significance of this statistic.

I continue to believe Tesla is forcing change on the industry. Back when AP 2.0 was introduced I posted a question - speculating that it would force the timetables of the competition to accelerate. Some folks disagreed - but only a couple months later MBZ appears to have finally thrown in the towel on their internally developed self driving software (their lab has been working on this for 20 years) and teamed up with Nvidia.

Folks at Volvo - who had an engineer famously call AP 1.0 an "unsupervised wannabe" - must be quietly stewing today now that Tesla has once again stolen the thunder from Volvo's raison d'etre - safety.

Nerds will argue the meaning of that stat on this thread but that isn't my point.

My point is the publicity of a highly dramatic NHTSA finding - 40% reduction in Tesla crash rates since Autopilot's introduction - will embolden other automakers to accelerate the introduction of their own autopilots. I also think it will embolden Tesla and its lawyers WRT the features of AP 2.0 and give them some public justification in releasing more and more features. The future of self driving is bright - this stat will certainly help give political cover to regulators to allow hands off driving with the full self driving suite in hardware 2 cars.

The stat certainly will bolster Musks' reputation and help cement his and Tesla's place in automotive history as true pioneers - not only in sustainable transport but in the biggest safety improvement in automobiles since the seatbelt. Again, to the naysayers - whether or not Musk's team invented unique technology isn't the point - the point is he was the one man with the cajones willing to actually be the first to deploy the tech and change the world.

40% is simply HUGE - especially for a first generation system that so many in the media publicly doubted and crapped on.

NHTSA’s full final investigation into Tesla’s Autopilot shows 40% crash rate reduction

 

[ecarfan]

Agreed. Elon's repeated assertions that AP cars are safer, and that it would be immoral to slow down or restrict Tesla AP deployment because people would die who would otherwise have been saved, have been vindicated by the NHTSA. It's full speed ahead on AP for sure.

 

[panjouser]

If you read through the entire study you will notice in footnote 22 "The crash rates are for all miles travelled before and after Autopilot installation and are not limited to actual Autopilot use."

That 40% reduction is just on cars with the hardware installed, it didn't even track the actual use. The assumption is that for actual miles driven with AP on vs. miles driven without AP engaged would be even safer than the 40% reduction of total accidents.

 

[thimel]

The same report says that AEB on all brands of cars reduces the accident rate by 40%.
For Tesla, the report doesn't differentiate between AP on or off as far as accident rate goes.
Seems likely the reduced accident rate in Tesla is due to AEB. There is certainly no information as to whether engaging AP helps or hurts. I wish there was.

 

[Gizmotoy]

The same report says that AEB on all brands of cars reduces the accident rate by 40%.
For Tesla, the report doesn't differentiate between AP on or off as far as accident rate goes.
Seems likely the reduced accident rate in Tesla is due to AEB. There is certainly no information as to whether engaging AP helps or hurts. I wish there was.

I was wondering about that myself. The report states, explicitly:

"The crash rates are for all miles travelled before and after Autopilot installation and are not limited to actual Autopilot use."

I'm trying to remember the feature timeline: were AEB and other safety features enabled before Autosteer, or simultaneously with Autosteer?

 

[GlmnAlyAirCar]

The same report says that AEB on all brands of cars reduces the accident rate by 40%.
For Tesla, the report doesn't differentiate between AP on or off as far as accident rate goes.
Seems likely the reduced accident rate in Tesla is due to AEB. There is certainly no information as to whether engaging AP helps or hurts. I wish there was.

If AEB alone caused the 40% reduction, this would be true across all models with AEB. However, this statistic is unique to Tesla, so it can not be attributed to AEB alone. Furthermore, AEB was released well before the general rollout of AP, so if this were the the reduction in accident rate would be coincident with the introduction of AEB in version 6.x, not general AP in 7.0.

Also, if this statistic compared the pre-AP accident rate with the accident rate only with AP active, this would have been spelled out in the report as it is not a direct apples to apples comparison. The only reasonable conclusion is that the comparison is all driving conditions pre- and post-autopilot.

 

[thimel]

If AEB alone caused the 40% reduction, this would be true across all models with AEB. However, this statistic is unique to Tesla, so it can not be attributed to AEB alone. Furthermore, AEB was released well before the general rollout of AP, so if this were the the reduction in accident rate would be coincident with the introduction of AEB in version 6.x, not general AP in 7.0.

Also, if this statistic compared the pre-AP accident rate with the accident rate only with AP active, this would have been spelled out in the report as it is not a direct apples to apples comparison. The only reasonable conclusion is that the comparison is all driving conditions pre- and post-autopilot.

Read the report again. In a different part of the report it states that AEB in all makes of cars reduces the accident rate by 40%.

 

[GlmnAlyAirCar]

Read the report again. In a different part of the report it states that AEB in all makes of cars reduces the accident rate by 40%.

Wrong. These are two different 40% statistics.

1. IIHS research shows that AEB systems meeting the commitment would reduce rear-end crashes [emphasis added]
by 40 percent.

2. The data show that the Tesla vehicles crash rate dropped by almost 40 percent after Autosteer installation.

(1) is a projection of AEB results. (2) specifically calls out autosteer in Teslas.

 

[u00mem9]

Wrong. These are two different 40% statistics.

1. IIHS research shows that AEB systems meeting the commitment would reduce rear-end crashes [emphasis added]
by 40 percent.

2. The data show that the Tesla vehicles crash rate dropped by almost 40 percent after Autosteer installation.

(1) is a projection of AEB results. (2) specifically calls out autosteer in Teslas.

Did Tesla cars have AEB prior to Autosteer?

 

[GlmnAlyAirCar]

Did Tesla cars have AEB prior to Autosteer?

Yes. It was available in version 6.x of the software. See post #6 above.

 

[u00mem9]

Since 'AEB' covers a wide range of capabilities, it would have been really great to show the reduction from just AEB (within Tesla data).

I get the sense that NHTSA is convinced this tech will lead to long term safety improvement (and I agree). But it would be refreshing if there could be objective evaluation rather than everything turning into advocacy.