Tesla on Autopilot crashes into police cruiser: Lansing, MI

[sq.in.]

FYI:

Tesla on autopilot crashes into state police patrol car in Lansing

LANSING, Mich. (WLNS) — Michigan State Police said a Tesla on autopilot drove into a Lansing area trooper’s patrol car. It happened around 1:10 on Wednesday morning as the trooper was i…

www.wlns.com

 

[CYBER_Y]

On the FSD beta or regular autopilot

 

[cobra]

'While driving with a suspended license"...

I think we know how this will go. Glad there are no injuries.

 

[boonedocks]

'While driving with a suspended license"...

I think we know how this will go. Glad there are no injuries.

Surely that will be spun as Tesla’s fault too

 

[powertoold]

It seems like the patrol car "failed to move over" as well. Not sure if it was necessary for the patrol car to be ~20% in the lane.

 

[designrs]

It’s really hard to ignore those flashing police lights at 1 AM, while driving on a suspended license.

 

[Dan D.]

It seems like the patrol car "failed to move over" as well. Not sure if it was necessary for the patrol car to be ~20% in the lane.

Whether they do it on purpose or not, yes it's a risk and a safety hazard, but has to be avoided regardless.

“Our officers are trained to park partially overlapped with the stopped vehicle in order to create a safety pocket. This is done so that the police car will be hit before an officer is. However, with the high speeds motorists are travelling, officers do get hit, and the results can be tragic."
https://globalnews.ca/news/3802816/...lls-for-local-drivers-to-slow-down-move-over/

 

[Dan D.]

It’s really hard to ignore those flashing police lights at 1 AM, while driving on a suspended license.

Do you think flashing lights can dazzle the AP cameras? All the strobes and flashing light bars certainly dazzle me.

 

[LowlyOilBurner]

And this is why you won’t be seeing a “button” anytime soon. People are too stupid to pay attention, apparently.

 

[MorrisonHiker]

And this is why you won’t be seeing a “button” anytime soon. People are too stupid to pay attention, apparently.

Exactly. They have the data and should be able to identify potential FSD beta candidates based on their Tesla driving history, years of ownership, etc., not on who can push a button and ignore the disclaimers.

 

[AlanSubie4Life]

It seems like the patrol car "failed to move over" as well. Not sure if it was necessary for the patrol car to be ~20% in the lane.

They do this for their own safety, as far as I understand it. If someone isn't going to notice a police car with lights on, sticking out in a lane by 20%, they're not going to notice anything, so may as well put the car there to slow them down some! I think that's the general idea, though maybe there is more to it. I think they angle the wheels and car so it has a tendency to fly into the roadway when hit, rather than into the disabled motorist and officer beyond the police car.

 

[powertoold]

They do this for their own safety, as far as I understand it. If someone isn't going to notice a police car with lights on, sticking out in a lane by 20%, they're not going to notice anything, so may as well put the car there to slow them down some! I think that's the general idea, though maybe there is more to it. I think they angle the wheels and car so it has a tendency to fly into the roadway when hit, rather than into the disabled motorist and officer beyond the police car.

Ya, I’m Not sure if the details of the stop are available. The other car that hit a deer should have fully pulled over. Again, I don’t know if it was necessary for the cop to be in the lane. I personally have never seen cops partially in the lane on highways (During stops). The only times they’re in the lane, they set up cones and/or flares.

 

[qdeathstar]

How do they know it was on autopilot?

 

[Cosmacelf]

How do they know it was on autopilot?

Hopefully the police asked the idiot driver and he told them. But even so, that doesn't mean it was actually on. Logs will show. Give it time.

That March 11 crash with a tractor trailer has now been deemed to NOT have been an autopilot crash:

U.S. investigating suspected Autopilot Tesla crash into police vehicle

The U.S. auto safety agency said on Wednesday it will send a team to investigate a crash involving a Tesla suspected of being in Autopilot mode when it struck a parked Michigan State Police patrol car.

www.reuters.com

 

[Dan D.]

That March 11 crash with a tractor trailer has now been deemed to NOT have been an autopilot crash:

https://www.reuters.com/article/us-...tesla-crash-into-police-vehicle-idUSKBN2B935M

The story says "Detroit police said Tuesday they do not believe that Autopilot was in use the March 11 crash based on 'all indications.'"

I'll wait for the report, or a consensus anyway. Assumptions can change. Here's the opinion from Tempe police on the Uber crash March 18, 2018. A fatal crash involving a self-driving Uber likely was "unavoidable" based on an initial police investigation and a review of video, Tempe Police Chief Sylvia Moir told The Arizona Republic. "It's very clear it would have been difficult to avoid this collision in any kind of mode (autonomous or human-driven)"

That was later changed to: Tempe police investigators released their findings that the crash was "entirely avoidable."

 

[AlanSubie4Life]

I'll wait for the report, or a consensus anyway. Assumptions can change.

I'm no Tesla FSD/AP evangelist, but I think you're showing your bias here. The Detroit situation you are referring to shows none of the hallmarks of an AP crash. Even if we had no witness & police testimony, AP/FSD would not be suspected, at all. It just doesn't make any sense, and it's not a likely failure mode of AP to miss a light and blow through an intersection at 20mph over the speed limit (or whatever it was - speed was clearly a factor). Nothing lines up!

For this stuff, we just have to call balls & strikes.

The accident at topic in this thread is totally different. It has all the hallmarks of an AP-driven crash. But the vehicle could also have been fully controlled by a human, and the driver may be lying about using AP. For this one we definitely have to wait to see what is discovered. In the end, though, I wouldn't be surprised if it ended up being AP failing to stop.

 

[Dan D.]

I'm no Tesla FSD/AP evangelist, but I think you're showing your bias here. The Detroit situation you are referring to shows none of the hallmarks of an AP crash. Even if we had no witness & police testimony, AP/FSD would not be suspected, at all. It just doesn't make any sense, and it's not a likely failure mode of AP to miss a light and blow through an intersection at 20mph over the speed limit (or whatever it was - speed was clearly a factor). Nothing lines up!

For this stuff, we just have to call balls & strikes.

The accident at topic in this thread is totally different. It has all the hallmarks of an AP-driven crash. But the vehicle could also have been fully controlled by a human, and the driver may be lying about using AP. For this one we definitely have to wait to see what is discovered. In the end, though, I wouldn't be surprised if it ended up being AP failing to stop.

Thanks, you've made a good point that the t-bone crash isn't likely AP, assuming the light was fully red and the truck was legally in the intersection. If the opposite was true then we would have still have had to consider AP.

It's not an anti-Tesla bias. I guess I'm just a stickler (some say nuisance) for statements written in the absolute sense.
"That March 11 crash ... has now been deemed to NOT have been an autopilot crash" vs "police said Tuesday they do not believe that Autopilot was in use the March 11 crash based on 'all indications.'"
Absolute vs Presumption

 

[brkaus]

It’s really hard to ignore those flashing police lights at 1 AM, while driving on a suspended license.

Perhaps drunk and napping?

 

[Barklikeadog]

Thanks, you've made a good point that the t-bone crash isn't likely AP, assuming the light was fully red and the truck was legally in the intersection. If the opposite was true then we would have still have had to consider AP.

It's not an anti-Tesla bias. I guess I'm just a stickler (some say nuisance) for statements written in the absolute sense.
"That March 11 crash ... has now been deemed to NOT have been an autopilot crash" vs "police said Tuesday they do not believe that Autopilot was in use the March 11 crash based on 'all indications.'"
Absolute vs Presumption

In use or not, where was the AEB in either case? It does not require driver attention nor is it in a testing phase. It is advertised as detecting an imminent collision, and reducing the severity, or avoiding a crash.

 

[AlanSubie4Life]

In use or not, where was the AEB in either case? It does not require driver attention nor is it in a testing phase. It is advertised as detecting an imminent collision, and reducing the severity, or avoiding a crash.

It is not guaranteed to detect stationary objects (the radar filters out stationary objects). It's probably much better at detecting objects that are still moving, but are being closed upon rapidly, since they can be distinguished from the background clutter. A vision-based system might do better if it can be made accurate and reliable enough.

The truck might have been perceived as moving, a bit, since the return would have been varying (truck was not there, then it was...but still stationary relative to the Tesla, since it was moving perpendicular to the Tesla's direction of travel). But probably not enough variation in the return. And I'm not sure it could do enough in that Detroit case anyway - the Tesla might have been moving at 65mph, for all we know, and there simply wouldn't have been much time to stop once the truck came into view from behind the building on the left.

The police car in this case is a perfect example of an object that wouldn't be flagged. It's stopped, so it's not there.

We all know that the Tesla WILL flag stationary objects, of course. Happens all the time with cars parked next to the road. But it’s not clear what circumstances are required for this to happen.