Regulators Open Investigation Into Fatal Crash in Tesla on Autopilot

[N-OS X-tasy!]

space-time wrote:
that being said, I would like to see autonomous cars being an order or magnitude safer than humans, as a minimum goal.

An order of magnitude is a VERY tall order.

[Ombligo]

I have a hard time believing that the driver could not have avoided the collision if he was watching the road. Alternate 27 is a straight shot divided roadway that runs for miles outside Williston (I live about 40 minutes away and have driven that roadway). There is nothing alongside the roadsides (except a gas station at that intersection). A driver could have braked, or run off the roadway into the wide shoulder or median.

He struck the trailer; tractor trailers are slow moving vehicles when turning. The Tesla had to have been a decent distance away when the truck started the turn, plenty of time for an alert driver to slow down or otherwise avoid the collision.

When the investigation is over, I'll bet the blame will boil down to faulty A.I. coupled with a driver who placed way too much trust in a new technology.

[RAMd®d]

I'm actually surprised a self driving car got something this simple so wrong.

This is why I asked about the kind of sensors used.

If optical, I'm not so surprised. If radar, I am surprised.


Maybe the Tesla brain KNEW about the truck and decided there was nothing that could be done, any other course of action would have resulted in more deaths.

Without knowing all, or at least more (as in the speeds), of the AI details, a human driver, not relying on auto pilot, I don't see how applying the brakes, even at the last second, would be worse than not applying them at all. I understand not turning or swerving, assuming the electronics even had the ability to weigh that option. But not braking at all? No way.

I think it more likely a failure of sensors, a failure of sensors to handle a particular situation, or a failure of programming to handle a particular situation.

In CA, and it most states, I'm guessing, the Tesla would absolutely have the right of way, with the truck required to yield to it. However, if the truck makes a turning movement, the Tesla would be required to yield it's right of way, if not doing so would cause a collision.

Many states' authorities list the suspect/at fault vehicle as Driver 1, V1, V01, etc. Not all states are the same, but the FHP's investigation might be declaring the rig at fault. NHTSA could change al that, or confirm it. Maybe.


Seems like a stretch.

I don't think so. Be that as it may, it's a standard investigative yard stick to help establish responsibility and possible cause.


Not only that, but since the "driver" had it in autopilot, he probably wasn't paying as much attention as a driver normally should.

Which to me seems to support the "not noticing the truck" supposition, at least until it was too late. Autopilot or no, I'm sure the driver would have tried to brake, if he had noticed the truck in time.


I wonder what a human driver could have done.

Don't we all. First, as mentioned, without the option of Autopilot, the driver presumably would have paid more attention. Depending on his distance to the truck, he should have seen the tractor before it started the turn. He should have seen it moving before the trailer was 'lost' in the sun. Again, this is supposition.

I think it possible, and maybe likely that the truck driver saw the Tesla, and assumed it had enough lead to start the turn while giving the car enough time to stop safely, even though he knew that if it didn't stop, there would be a collision. But who doesn't stop?

When starting his turn, did the driver allow enough time for the Tesla to stop safely? Did the driver see it before turning? Was there enough time for the Tesla driver to stop safely? Autopilot or no, I think at this point in legislation and technology, the driver still has the ultimate responsibility. I don't for a second believe the driver has a reasonable expectation to absolutely trust the autopilot without oversight.

[Mike Johnson]

NewtonMP2100 wrote:
was watching a Harry Potter video

Frank Baressi, 62, the driver of the truck and owner of Okemah Express, said the Tesla driver was “playing Harry Potter on the TV screen” at the time of the crash

The movie “was still playing when he died,” Baressi told The Associated Press

Tesla Motors said it is not possible to watch videos on the Model S touch screen. There was no reference to the movie in initial police reports.

The investigators found a portable DVD player at the scene. http://gizmodo.com/it-looks-like-the-tes...1782980532

[M A V I C]

Ombligo wrote:
I have a hard time believing that the driver could not have avoided the collision if he was watching the road. Alternate 27 is a straight shot divided roadway that runs for miles outside Williston (I live about 40 minutes away and have driven that roadway). There is nothing alongside the roadsides (except a gas station at that intersection). A driver could have braked, or run off the roadway into the wide shoulder or median.

He struck the trailer; tractor trailers are slow moving vehicles when turning. The Tesla had to have been a decent distance away when the truck started the turn, plenty of time for an alert driver to slow down or otherwise avoid the collision.

I brought it up on satellite view. What the drawing doesn't show is the massive median. It also doesn't show that the intersection is not perpendicular. There's no possible way a semi truck could have made a left turn that sharp, so fast that the Tesla couldn't have stopped. Any decent driver would have seen the semi starting the turn and then been ready to pounce on the brake if the cab of the truck went into the oncoming lanes. There's no way for the cab of the semi to have cleared the intersection without a competent driver having seen it begin to turn, enter the median area and enter the lanes of traffic - all opportunities to slow down. Measuring it out, the semi's intentions were clear for about 150' (if not more). If the semi managed to make that corner at 35mph (most likely impossible) the Tesla still had three seconds to respond. At 20mph, 5 seconds.

At 60mph, the Tesla was covering 88 feet per second. So in the 3-5 seconds the truck was turning (if not much longer) the Tesla covered 260-440 feet. From what I can tell, it needs less than 150' to come to a complete stop. Obviously the experts will research all this and be using something better than Google Earth to figure out the distances, but it sure seems like that Tesla had more than enough time to come to a complete stop.

[max]

M A V I C wrote:


Looks like the Tesla had the right of way.

Not necessarily, trucks that size quite often require a bit of work to turn safely in the urban environment.
This truck could have initiated the turn when there was no oncoming visible traffic and he had right of way....

[max]

Another aspect....

Mobileye, the Israeli company that makes some of Tesla’s autopilot technology, said in a statement that the crash involved a vehicle turning in front of the Tesla. It said current automatic emergency braking system isn’t designed to act in such situations. Mobileye said its systems will include lateral turn detection in 2018.

http://ktla.com/2016/07/03/tesla-autopil...-tells-ap/