Fatal autopilot crash, NHTSA investigating...

[DrPhoton]

Do you get an alert when the cam looses lane markers?

The Tesla AP system gives very good feedback to the driver. It clearly indicates on the dash display whether it is sensing one or both lane markers. It also indicates whether it is sensing the presence of an additional lane to the left and/or the right. If the system completely loses lane markers on both sides, I'm assuming it will sound an alert tone as well as indicating this on the dash display, although I've never encountered this situation myself.

 

[Tam]

Ok, the area where we see the bikes on the streets are 35mph. The signs tell them to take the entire lane. What is scary for us, is following a car at a safe distance, car suddenly changes to left lane exposing slow bikes and not much time to react.

I view autopilot like cruise control and was concerned when CEO said no touch between cities. Im concerned people would be distracted and focus on other things unlike cruise control.

Interesting times

I believe current Autopilot design is:

1) simple scenario of freeway speed driving.
2) simple scenario of ultraslow speed for summoning and parking.

When you talk about 35 MPH, that's is like city driving which is the most difficult stage to implement.

Current autopilot is not to be used in city driving because it is not yet designed for.

 

[Raffy.Roma]

So can anyone speak authoritatively on the respective roles of camera and radar here?

Actually the radar alone without the need of the camera should manage to detect anything in front of the car. On the contrary from what I heard also the camera works in the autopilot system to detect things in front of the car. This is wrong in my opinion because a camera can make mistakes in detecting things as it happened in this case. Don't mean that Tesla autopilot engineers are wrong. I only would like to say my opinion.
Then on the matter of difference between autopilot and autonomous driving I would like to say that when a Tesla owner has an autopilot feature working so good as the Tesla autopilot the Tesla owner starts thinking of having an autonomous driving feature. It's normal. So I think that Tesla should change the following AP2 in AD2 (Autonomous Driving) and produce a new system guaranteeing that this kind of accidents will never happen anymore.

 

[DrPhoton]

...My GUESS is, he was watching the movie on a phone or tablet and runing the sound through thespeakers via BT since the sound from a phone or tablet is pretty weak.
.

Reading all these threads, it seems like many people already drawn the conclusions that:

1. The driver was watching a movie because the truck driver who caused the accident "heard" it and a portable DVD player was found among the wreckage
2. "Eye witness report" from a very questionable (and sensational) local news station report which stated that the car was travelling much faster than 85mph while AP limits to 90mph. (BTW, could the statement from this "witness" become self-incriminating evidence for speeding?)

I just want to point out that these two "facts" were never quoted in any police account of the accident except a DVD player was found. (please correct me if I am wrong with the official source) The funniest quote I got from that local ABC news clip was the home owner who became an AP expert when he was quoted that AP must have avoided the two trees but failed to avoid the phone pole. I guess random chances never occurred to the news crew. In fact this home owner, who most likely have never driven a Tesla with AP, was given a lot of time to express his opinion (not facts) on AP. If there is a Rotten Tomato award for news cast, I would think that this news crew should be among the finalists.

When I ran into the "Inside Edition" video regarding this accident on YouTube, I immediately smelled something...

As for myself, I am eagerly waiting for the final investigation result from NHTSA and FHP to put this to rest.

I ran across this comment about the Tesla driver from one of his close friends that makes it seem likely that if there was audio playing in the car after it came to rest, it was most likely a movie sound track:
Tesla elaborates on Autopilot’s automatic emergency braking - Driver's Friend's Comment

 

[omarsultan]

f the system completely loses lane markers on both sides, I'm assuming it will sound an alert tone as well as indicating this on the dash display, although I've never encountered this situation myself.

The autosteer icon goes from blue to grey, you get an audible alert and a message on the IC to put your hands on the steering wheel.

 

[KZKZ]

I'm curious why Tesla is focusing only on the trailer in the accident and not the truck and the trailer.

It's not like the trailer was backing into traffic, and the trailer was not the first obstacle for the AP to see (or should have seen).

The AP would have seen the truck portion first.

The truck was directly in the Tesla's line of travel first and should have been registered on the radar. The truck is certainly low enough, it has big tires and it was moving.

If nothing else, I would have expected for at least the collision warning system to activate based on the truck portion of the big rig.

Thoughts on why Tesla's focus is only on the trailer?

 

[JeffK]

I'm curious why Tesla is focusing only on the trailer in the accident and not the truck and the trailer.

It's not like the trailer was backing into traffic, and the trailer was not the first obstacle for the AP to see (or should have seen).

The AP would have seen the truck portion first.

The truck was directly in the Tesla's line of travel first and should have been registered on the radar. The truck is certainly low enough, it has big tires and it was moving.

If nothing else, I would have expected for at least the collision warning system to activate based on the truck portion of the big rig.

Thoughts on why Tesla's focus is only on the trailer?

I feel like a broken record. The truck is moving laterally which is outside the current autopilot spec. The cab was not in front of the car at the time of impact so why would it stop for what might appear to be a transient unknown obstacle at a distance that's there one second and gone the next. The trailer is high and the radar system ignored it but more importantly it's a stationary object that form the point of view of the sensors appeared from nowhere in the middle of the road. Those systems from all manufacturers don't do well with stationary objects, especially ones that don't look like vehicles.

 

[nikielizabeth]

I guess being consistent isn't your strong point.



Whenever an author posts anything on the Internet, he or she should reasonably expect that it will be read, downloaded, printed out, forwarded, and even used as the basis for other works to some degree. So, just by posting, an author impliedly grants a limited license to use her work in this manner.

Isn't the 'imply' granted to the actual site host and not the trolling blogger w/ nothing better to do?

 

[xav-]

Anybody thinking Brown could have voluntarily tried to pass below the truck? A la fast and furious?

 

[MP3Mike]

Anybody thinking Brown could have voluntarily tried to pass below the truck? A la fast and furious?

I thought about that, but you would have thought that he would have tried to lay down sideways or something, and that would be obvious from the injuries and I would think that would have been reported. But maybe not.

 

[JeffK]

Anybody thinking Brown could have voluntarily tried to pass below the truck? A la fast and furious?

That will show up in the logs if the owner pressed the accelerator pedal so there is a way for investigators to know for sure.

I've had too many close calls and a serious accident with semi-trucks failing to yield. I don't trust them at all and wouldn't even try it.

 

[brucet999]

Ask yourself: Would the driver have driven at > 85mph while being so distracted if he didn't have AP?

Correct me.please, if I'm mistaken, but I think Auto Pilot speed is limited to 85 mph. If so, with AP on, he could not have passed someone who was driving 85mph.

 

[brucet999]

It's limited to 90 MPH. But if her account is accurate it would seem there would need to be more than a 5 MPH delta between their speeds.

Mike

That bothered me too, but he could have been going 90, accelerated as he overtook her, then returned to AP mode.

 

[brucet999]

I haven't read the complete thread, so I'm sorry if this has been brought up before, but I'm quite annoyed by this part of Tesla's statement:

This is the first known fatality in just over 130 million miles where Autopilot was activated. Among all vehicles in the US, there is a fatality every 94 million miles. Worldwide, there is a fatality approximately every 60 million miles.

I mean, how are these figures even remotely comparable? Are they trying to tell us that the AP controlling a very safe car under very benign driving conditions is safer than a human driver who is e.g. driving on a twisting country road in pouring rain and even more safe than e.g. a human driver who drives an ancient car on a washed out road high up in the Andes?
Only miles driven by human drivers in equally benign conditions that allow the AP to operate should be used as a standard to measure the safety of a human driver vs. AP.

I made the same observation. More germane would be the ratios of non-AP collisions of Model S (at speeds within AP range) vs Model S with AP in operation.

 

[mrElbe]

Anybody thinking Brown could have voluntarily tried to pass below the truck? A la fast and furious?

And misjudged the height of the gap from road to trailer.

 

[brucet999]

Tesla's direction and approach before impact (straight away, no major elevation changes/cresting) - Semi pulls in front from left at cross section going right into BP side street. (edited for correction)

Tractor trailer turns left in front of Tesla into BP street (correction from original post)

Birdview - Red (X) impact location - Tesla traveling southeast comes to rest near box -

Thanks for posting the pic. This tells me that the tractor trailer could not have been going fast when he started his turn since it was a sharper angle (maybe 100 degrees) than the normal 90 degree left turn. Truck driving lesson video posted earlier of how to make a left turn on normal 4-lane streets without median, showed it took 12 seconds to complete the turn. In this case it is a sharper turn, plus 50-foot median strip plus 30 feet of eastbound pavement, so it would have taken the tractor even longer to get to where only the trailer blocked the right lane; perhaps something like 15 seconds.

The 65 mph stated by FHP would place the Tesla just over 1/4 mile away at the start of the tractor-trailer's left turn. If the witness who said that she was going 85 when passed by the Tesla is correct, then he would have been going 90 (limit of AP) and would therefor have been 3/8 mile away at start of left turn maneuver.

Did the truck driver assume a slower speed and so conclude the 3/8 mile distance would be enough for him to clear the road?
Did he assume the Tesla would slow down?
Did he simply fail to see it?

 

[182RG]

If the witness who said that she was going 85 when passed by the Tesla is correct, then he would have been going 90 (limit of AP) and would therefor have been 3/8 mile away at start of left turn maneuver.

Did the truck driver assume a slower speed and so conclude the 3/8 mile distance would be enough for him to clear the road?
Did he assume the Tesla would slow down?
Did he simply fail to see it?

All of us gauge closure rates for other vehicles based on experience. How many of us take more than a split second glace at approaching traffic and decide to act based on those experiences and speeds we are accustomed to seeing. A car traveling 90 is going to close the distance a lot quicker than expected.

If facts prove out that the Tesla was traveling at that rate of speed, you simply cannot hold the truck driver accountable for any part of this accident.

 

[wdolson]

The folks who have suggested that Tesla should modify their AEB and/or TACC so that it is more cautious and potentially results in more false positives leading to slowing or braking when it's not needed are not thinking it through. If the system did this it would not only become dangerous, but it would be unusable and therefore pointless. With TACC the way it is now, it occasionally misjudges which lane vehicles in front of it are in, especially on curves. As a result, our Model S will occasionally slow when overtaking a vehicle in an adjacent lane. In at least one case, this has almost resulted in a road rage incident: a clearly impatient driver pulled into the fast lane close behind me while we were entering a curve and almost simultaneously my car rapidly slowed (deceleration was sufficient to cause my brake lights to illuminate). The driver behind me assumed that it was a jerk move on my part (actually a logical assumption given the situation) and started tail-gating me and swerving aggressively (not a logical reaction, but unfortunately not a rare one either). I quickly pulled aside to let him pass to avoid aggravating him any further, but you can easily see where this could have gone. It's not outside the realm of possibility to imagine that unnecessary braking or deceleration could lead to violent acts by other drivers who could misinterpret the reason for the maneuver.

I usually drive with TACC engaged, and often with AP operating, but I've begun to be able to anticipate when it may incorrectly slow while overtaking other vehicles and I compensate by applying a bit of pressure to the accelerator in order to counteract the false detection.

In defense of the TACC's difficulty in determining when or if to slow when overtaking other vehicles, it is also fairly common for human drivers to slow when vehicles in the lane next to them slow down, so it's not necessarily dangerous - just annoying. What's different about TACC is that for other drivers around me its behavior is different than what's normal for human drivers, leading to possible misunderstandings.

If the AEB system were modified so that it had even one false braking event, I think some owners would be so frightened/angry that they would disable it completely. If the AEB were so sensitive that it happened as often as once every few months, it would probably cross the threshold from useful to hated for most people. IMHO, drivers will be MUCH less tolerant of false AEB triggering than many posters seem to believe.

The general public is ignorant of regen braking and how the brake lights work when it's engaged. I recall a few months back Carscoops.com posted a video of what they said was a Model S road rage incident where a Model S pulled in front of a truck and the brake lights started to flash. Someone in the comments pointed out that was the regen braking system kicking in and had nothing to do with road rage. Those guys are usually knowledgeable about most things Tesla, but even they thought it was road rage and not the car doing its thing.

I wonder how long it's going to be until a Tesla driver gets shot for using regen?

 

[ItsNotAboutTheMoney]

58 pages already. Just think what'll happen when the NHTSA actually publicizes its findings.

 

[wdolson]

I'm curious why Tesla is focusing only on the trailer in the accident and not the truck and the trailer.

It's not like the trailer was backing into traffic, and the trailer was not the first obstacle for the AP to see (or should have seen).

The AP would have seen the truck portion first.

The truck was directly in the Tesla's line of travel first and should have been registered on the radar. The truck is certainly low enough, it has big tires and it was moving.

If nothing else, I would have expected for at least the collision warning system to activate based on the truck portion of the big rig.

Thoughts on why Tesla's focus is only on the trailer?

Teslas have doppler radar which senses how fast objects are going with respect to you in the same direction of travel you're traveling in. If something is moving in your direction, but at an angle, it's going to get an inaccurate speed. It can't detect if something perpendicular to you is moving or not. Something moving at a 90 degree angle to you will be seen as stationary.

I thought about that, but you would have thought that he would have tried to lay down sideways or something, and that would be obvious from the injuries and I would think that would have been reported. But maybe not.

He may have, but from the pictures of the damage, it may not have saved him. The damage to that car extended down a fair ways.