My friend's model X crashed using AP yesterday

[Tam]

...Is it possible that he engaged TACC instead of autopilot which would explain why the car maintained speed but did not stay on the road....

I don't think that is the issue here. Tesla stated:

"no force was detected on the steering wheel for over 2 minutes after autosteer was engaged (even a very small amount of force, such as one hand resting on the wheel, will be detected). "

Tesla clearly stated that "Autosteer" was in effect.

However, although AutoSteer is very good, much better than others on the market, but it is certainly not perfect. That is why the owner manual requires hands-on driving.

And if you must do hands-off driving, choose the one the manual recommends: divided median highways.

And if you do it on no-median roads, do it slowly, very slowly.

And if you do it on a mountainous road, make sure your hands on the steering wheel to take over at any time.

 

[hill]

The minute i saw "the friend" urgently come into a public forum to tell us what happened, I knew there was something fishy and they would end up blaming Tesla for something that is clearly their fault.

What - you don't think his FRIEND (who likely told him to engage AP - on a winding road with faded centerlines & soft shoulder) has Tesla's deep pockets?
.

 

[imherkimer]

If you are looking on Google Earth and follow MT State Route 2 south from I-90 a few miles, the road passes a narrow area between a steep ridge on the left and on the right are the railroad tracks, and then the Jefferson River. Here is the kind of guard "rail" described by the OP. It is visible in the area where the road, rail line, and Jefferson river seem to converge. This area seems to fit the description.

I would not use Autopilot at night on this stretch of road. There is a straight stretch and then at this point road begins to curve between river and ridge.

One thing that puzzles me is this: if the driver was going to West Yellowstone from Whitehall, at night, why is he on Route 2 and not on Route 359, which is much straighter, more direct to the destination, and a better road?

 

[legoman]

Not defending the driver at all, but I thought it was interesting he claimed to have used autosteer for 600km earlier in the trip. So he wasn't an inexperienced user of the system. He should have been familiar by that point as to when the steering is less perfect and needs more attention.

This is one reason why I think independent access to, and review of, log data would be interesting. The logs should show either 1) loss of lane markings coupled with no input to steering (i.e. the car kept driving straight and didn't "see" that the road curved left; OR 2) that there was an autosteer input to the right either because it thought it "saw" the road continue right OR because it lost track of the lane.

Second point of interest is if or why the car continued forward. We know from the photos there was no airbag deployment, which is questionable with an impact severe enough to rip the front wheel off, but perhaps understandable if the point of impact was narrow enough (the post) to shear pieces from the car without transferring enough blunt force to the front bumper to trigger the airbags. But aren't there some connected systems involved? Wouldn't the car have either registered the catastrophic loss of hydraulics to the front wheel brake? Or registered the sudden change in steering (surely the force of ripping a wheel off moves the steering mechanism at least a bit)? Either could have been an indicator of an accident occurring and to immediately disengage the drive. So digging deeper into the logs would be enlightening.

I'm a fan of autopilot and I want it to continue to learn and improve. I hope to never encounter an emergency driving situation, but if I do I want the car to have learned from all prior accidents to perform optimally.

 

[atr2016]

If AP is being used as a true assist (driver remains 100% engaged), 5 mph below the speed limit (because it's late & dark), isn't it safer than 100% manual driving on these so-called "known unsafe roads"?

Too many stories focusing on distracted/sleepy/drunk/whatever drivers, not applicable to most people.

 

[CmdrThor]

If AP is being used as a true assist (driver remains 100% engaged), 5 mph below the speed limit (because it's late & dark), isn't it safer than 100% manual driving on these so-called "known unsafe roads"?

Too many stories focusing on distracted/sleepy/drunk/whatever drivers, not applicable to most people.

Maybe but that is irrelevant to this thread because the driver did not have his hands on the wheel and he was exceeding the speed limit.

 

[atr2016]

Maybe but that is irrelevant to this thread because the driver did not have his hands on the wheel and he was exceeding the speed limit.

Irrelevant until people start suggesting disabling the feature under those conditions.

 

[sillydriver]

I've just skimmed through this whole thread and I'm still confused about the location. Montana state 2 comes off I90 at exit 256 together with state 359 (probably the most likely location) but it also comes off earlier (further west) in Whitehall and even earlier in Butte. The phone pictures imply Whitehall, but if that is where the car was towed, it could have been any of the exits. Getting off at exit 256 would get you to federal 287 and ultimately Yellowstone, but staying on state 359 at that exit would get to federal 287 much more directly, so it's a strange route. It might have been Butte because I think they would have had to stop at the supercharger on state 2 right off of I90 in Butte. If they were then going to stay on 2 from Butte to 41 to 287, they would have hit a short section of hairpin (no kidding!) turns several miles further on 2. If the accident was due to plowing into those hairpins at full speed it puts the whole thing in a different light.

EDIT - but I don't see a guard rail at the hairpins on street view.

I wonder if there is an accessible public record of where the accident actually occurred?

 

[JohnSnowNW]

I've just skimmed through this whole thread and I'm still confused about the location. Montana state 2 comes off I90 at exit 256 together with state 359 (probably the most likely location) but it also comes off earlier (further west) in Whitehall and even earlier in Butte. The phone pictures imply Whitehall, but if that is where the car was towed, it could have been any of the exits. Getting off at exit 256 would get you to federal 287 and ultimately Yellowstone, but staying on state 359 at that exit would get to federal 287 much more directly, so it's a strange route. It might have been Butte because I think they would have had to stop at the supercharger on state 2 right off of I90 in Butte. If they were then going to stay on 2 from Butte to 41 to 287, they would have hit a short section of hairpin (no kidding!) turns several miles further on 2. If the accident was due to plowing into those hairpins at full speed it puts the whole thing in a different light.

EDIT - but I don't see a guard rail at the hairpins on street view.

I wonder if there is an accessible public record of where the accident actually occurred?

Google Maps

 

[sillydriver]

No luck. There is a searchable national database of accident reports but those all involve death or injury. The Montana highway patrol has a database of traffic incident reports - minor incidents included - but it's only searchable on their website for incidents in the last ten days.

 

[sillydriver]

Google Maps

So it's near Whitehall. Odd route / location given where I think they were trying to go.

 

[CuriousG]

We noticed that the barrier posts is about 2 feet from the white line. The other side of the barrier is a 50 feet drop, with a railroad at the bottom, and a river next. If the car rolled down the steep slope, it would be really bad.

Google Maps

I see the railroad tracks but where's the river?

 

[JohnSnowNW]

I see the railroad tracks but where's the river?

Just up the road.

 

[CuriousG]

No luck. There is a searchable national database of accident reports but those all involve death or injury. The Montana highway patrol has a database of traffic incident reports - minor incidents included - but it's only searchable on their website for incidents in the last ten days.

Perhaps someone could call Montana Highway Patrol and ask if they have an exact location of the accident. They're probably all aware it's exact location and highly publicized now.

 

[imherkimer]

I see the railroad tracks but where's the river?

Follow route 2 south as it parallels route 359. After a straight stretch which is open, the road, railroad, and river converge near a ridge that appears to the left. On that curve, where river, rail, and road are closest together, you can see the "guardrail" appear on the right.

 

[sillydriver]

I see the railroad tracks but where's the river?

That description implies they got off I90 at exit 256, looking at google maps. The railroad is there. The river is the Jefferson river.

Google Maps

 

[imherkimer]

Yes, following south on 2 along the Jefferson river, a few miles after it turns south away from I90

 

[CuriousG]

Just up the road.

The only thing that's remotely close that crosses the railroad tracks and water would be labeled ditch or creek.

I'm referring to where the accident happened. If the "river" is a few miles out or passed, how would that be relevant from the victim's description?

 

[JohnSnowNW]

So, are we talking this area, now:

Google Maps

The victims description is confusing/ has changed...is the issue.

 

[imherkimer]

So, are we talking this area, now:

Google Maps

The victims description is confusing/ has changed...is the issue.

Yes, and I agree, it doesn't stack up with the "victim's" statement as to where the accident occurred, but it is the only place that truly fits the description, and the photos.