Fatal autopilot crash, NHTSA investigating...

[Garlan Garner]

None of us have purchased that hardware yet.

Yes, you are correct. However everyone has purchased hardware that has had its software upgraded and improved.

The point is this.....Tesla provides free updates as well as some paid upgrades associated with their hardware - which is fantastic.

A person is a lot less likely to turn off hardware that will receive software updates.

 

[Tam]

....Tesla provides free updates as well as some paid upgrades associated with their hardware - which is fantastic...

Fantastic indeed!

Mine is still a classic without Autopilot but I also received some benefits from Autopilot firmware updates as well.

Mine can now automatically close my garage door when I drive away and open it automatically when I come back home.

That's a feature from summoning but since a classic can't do summon, at least it still got a consolation prize

 

[dhanson865]

That doesn't tell me anything I don't already know.

Is Sterling Anderson mistaken when he says autopilot even in manual mode can prevent an accident where someone steps into the street? If he's referring to the AEB, I'm not sure if hitting the pedestrian at 35 mph instead of 60 mph can really be called a save. (Of course, it certainly better than hitting the pedestrian at 60 mph!)

It might be a save if it is the difference between hitting a pedestrian at 60 mph or getting to the spot where they were a second or two later at 35 mph giving them time to get out of the way. They might only need one step to clear the path of the car.

 

[Garlan Garner]

It might be a save if it is the difference between hitting a pedestrian at 60 mph or getting to the spot where they were a second or two later at 35 mph giving them time to get out of the way. They might only need one step to clear the path of the car.

So True.

 

[Max*]

It might be a save if it is the difference between hitting a pedestrian at 60 mph or getting to the spot where they were a second or two later at 35 mph giving them time to get out of the way. They might only need one step to clear the path of the car.

It's not only that, there's a reason most residential neighborhoods have a speed limit of 25mph. At 25mph, an impact with a pedestrian has a low mortality rate (~15%). The higher the speed limit above 25mph, the higher the mortality rate.
So the mortality rate of a pedestrain at 60mph are ~55%, the mortality rate at 35mph is only ~35%. I'd take a 2/3 chance of living over a 1/2 chance of living.

In addition, the risk of severe injury at 25mph is about 25%. The risk at 35mph is ~65%. The risk at 60mph is close to 100%.

Source: https://www.aaafoundation.org/sites/default/files/2011PedestrianRiskVsSpeed.pdf (there's a ton of research, most of them have roughly the same numbers)

/side rant

 

[Tam]

Tesla Looking At Cameras, Radar In Deadly Autopilot Crash

Un-named source: A private briefing with U.S. Senate Commerce Committee staff members yesterday, Thursday.

2 main theories were cited:

1) Both Camera and radar failed to detect a crossing tractor-trailer.
2) Camera failed to detect and the radar signal was interpreted as false alert such as an overpass or sign.

 

[Garlan Garner]

Tesla Looking At Cameras, Radar In Deadly Autopilot Crash

Un-named source: A private briefing with U.S. Senate Commerce Committee staff members yesterday, Thursday.

2 main theories were cited:

1) Both Camera and radar failed to detect a crossing tractor-trailer.
2) Camera failed to detect and the radar signal was interpreted as false alert such as an overpass or sign.

Shame on them for publishing an "un-named" source.
I always associate unnamed sources with unsubstantiated data.

 

[X Fan]

Tesla Looking At Cameras, Radar In Deadly Autopilot Crash

Un-named source: A private briefing with U.S. Senate Commerce Committee staff members yesterday, Thursday.

2 main theories were cited:

1) Both Camera and radar failed to detect a crossing tractor-trailer.
2) Camera failed to detect and the radar signal was interpreted as false alert such as an overpass or sign.

Government at its finest.....

 

[artsci]

Tesla Looking At Cameras, Radar In Deadly Autopilot Crash

Un-named source:

Often means mere speculation or totally made up.

 

[Tam]

http://www.nytimes.com/2016/07/30/b...es-but-not-autopilot-in-fatal-crash.html?_r=0

In addition to 2 above theories, it also reports that:

1) Autopilot technology was not at fault
2) The braking systems as “separate and distinct” from Autopilot
3) "some deaths might occur as automakers developed and perfected these kinds of innovations, the safety benefits outweighed the risks"
4) Kelley Blue Book analyst: "The reliability of Autopilot is a sensitive matter for Tesla because the company tries to present itself to customers and investors as a technology company rather than an automaker."

The anonymity of a Senate staff member might be well kept but how big is the room? Can you do a process of elimination to identify the unnamed staff?

 

[rscott0]

http://www.nytimes.com/2016/07/30/b...es-but-not-autopilot-in-fatal-crash.html?_r=0

The article makes it sound fishy that Tesla considers automatic braking separate from AP. But check out Tesla's marketing of AP, and it doesn't mention anything about automatic braking. And Consumer Reports (see Reality Check: It Wasn't Tesla's Autopilot That Prevented Crash in Viral Video ) says that auto braking is distinct from AP.

 

[traxila]

The article makes it sound fishy that Tesla considers automatic braking separate from AP. But check out Tesla's marketing of AP, and it doesn't mention anything about automatic braking. And Consumer Reports (see Reality Check: It Wasn't Tesla's Autopilot That Prevented Crash in Viral Video ) says that auto braking is distinct from AP.

All of the mainstream rags have it in for Tesla but NYT in particular since Tesla basically outed their Auto Reporter as a liar doing a hit piece when he killed the battery and put up an article with a picture of the car being towed. So yeah, they are going to try and spin negative every time. It not like Tesla will pull their ads or anything. Lol.

 

[zambono]

Tesla is faulting brakes on crash
http://mobile.nytimes.com/2016/07/3...-brakes-but-not-autopilot-in-fatal-crash.html

 

[Electroman]

This kind of slanted and cherry picking facts and exaggeration is par for all major rags like NYT, WSJ and LAT.

There was no emphasis and elaboration to the fact that Brown was driving 10 miles over the speed limit while not paying attention on a road that has cross traffic. All of those three conditions together is the root cause.

I say hit them where it hurts them. Let Tesla and its dealers pull all the ads from NYT. That will teach them a lesson.
-

 

[Az_Rael]

say hit them where it hurts them. Let Tesla and its dealers pull all the ads from NYT. That will teach them a lesson.

Does Tesla buy ads in magazines/newspaper? I thought their marketing strategy was more grass roots and word of mouth.

 

[traxila]

Does Tesla buy ads in magazines/newspaper? I thought their marketing strategy was more grass roots and word of mouth.

No they do not. We are being funny and also pointing where the rags' loyalties lie. Status quo, ad dollars, petro economy etc

 

[ecarfan]

I say hit them where it hurts them. Let Tesla and its dealers pull all the ads from NYT. That will teach them a lesson.
-

Tesla does not advertise in either print or digital media.

Tesla does not have "dealers". All Tesla Showrooms are owned by Tesa.

The NYT piece failed to state the obvious: the Tesla clearly states the driver has full responsibility for controlling the car and, if using AP, for taking over at any time if conditions warrant. It appears very likely that in the Florida crash the driver ignored the road while using AP, and the truck driver made an unsafe left turn. Fatal combination.

 

[Tam]

The NYT piece failed to state the obvious: the Tesla clearly states the driver has full responsibility for controlling the car and, if using AP, for taking over at any time if conditions warrant. It appears very likely that in the Florida crash the driver ignored the road while using AP, and the truck driver made an unsafe left turn. Fatal combination.

I agree that there is a violation of the terms of use in this case (driver agrees to maintain control and responsibility whenever a driver activates the feature.)

However, it is also important to understand the technical aspect of what are the current limits of Autopilot so drivers can be informed and engineers can improve the system.

Reports have been focusing on the separation of the Automatic Emergency Brake system and Autopilot.

Tesla Blames Automatic Braking System for Fatal Crash

Thus, it might be important to understand the difference.

Autopilot: "controls your steering, changes lane, and adjusts the speed of your car only."

On its generic webpage, Automatic Emergency Brake system is not listed in Autopilot descriptions/features.

On its Design webpage. Autopilot is not listed as "Safety" option. It is listed as "Convenience" option. It does not list Automatic Emergency Brake system as its subset. Instead, Automatic Emergency Braking system is listed as standard Equipment under "Safety."

Ok, now we know the difference.

So if the AutoSteer was functioning fine, did the log confirm the truck driver's statement that the Model S changed lane from left (fast) lane to right (slow) lane?

Regardless of which of two theories is true, what modifications can be made to make sure that won't happen again?

Mobileye says it will have Lateral Turn Across Path (LTAP) detection capabilities beginning in 2018 but not now but importantly it will not support Tesla beyond current chip version.

Elon says he works with Bosch to improve radar.

With an addition of LIDAR, proponents says this scenario is perfect for detection especially when the obstacle is white (good LIDAR reflection for better detection) and the weather is good with no rain, snow, fog, dust storms... Should Tesla seek a feasibility to add LIDAR or should the rest of the rest of the industry be encouraged to drop LIDAR and go with current Tesla's non-LIDAR direction?

 

[KZKZ]

New accident with autopilot-equipped car: http://www.dt.no/ulykke/trafikk/nyhet/flere-biler-og-mc-involvert-i-trafikkulykke/s/5-57-409504

(Camera by the rear mirror means it's new-ish, with autopilot hardware.)

It was a chain collision involving three cars and one motorcycle. Looks like the Model S was the last vehicle in the chain, and failed to stop. The motorcyclist is in the hospital while the other drivers were unharmed. This is a divided motorway just after a 62.5 mph (100 km/h) zone turns into a 56.25 mph (90 km/h) zone.

Whether or not the Tesla owner had paid for autopilot activation, it should be running in the background. And whether or not the driver had enganged autopilot, the autobraking should activate.

The way Tesla implemented AEB leaves a lot to be desired. It's really just a partial AEB system as it will only slow the vehicle by 25mph, and worse, it stops working as soon as the driver hits the brake pedal, even if the driver doesn't brake forcefully enough to avoid an accident.

Other manufacturer's AEB systems will brake all the way to a stop, and will also continue to brake the appropriate amount even if the driver hits the brake too slowly or too softly.

 

[JohnSnowNW]

The way Tesla implemented AEB leaves a lot to be desired. It's really just a partial AEB system as it will only slow the vehicle by 25mph, and worse, it stops working as soon as the driver hits the brake pedal, even if the driver doesn't brake forcefully enough to avoid an accident.

Other manufacturer's AEB systems will brake all the way to a stop, and will also continue to brake the appropriate amount even if the driver hits the brake too slowly or too softly.

They are not the only company that has AEB systems that won't bring the car to a dead stop with speeds above 30 mph,or the only one that will cancel if the driver hits the brake...so try to hide your negative Tesla slant a bit.