Can Tesla Data Help Us Understand Car Crashes?

[L_Mont]

This is today in https://www.nytimes.com/
Video : https://vp.nyt.com/video/2022/08/16/101991_1_tesla-crash-hp-promo_wg_720p.mp4

We had a similar situation with our Model Y. While in Self Driving on I95 in South Carolina, the car braked hard because there was a car stopped on the left grass median. Luckily all the cars behind us were able to brake and avoided an accident.

 

[Silicon Desert]

Well, from what I see, this was not a situation where the car braked hard. It seemed to slow down a little yet basically maintained about the same speed. Not a hard braking incident. The problem here seems to be that the car behind the Tesla did not anticipate the Tesla staying in the lane after partially moving over the line. He was too aggressive. It would have been nice to see the car behind the Tesla for a longer period of time than just one second before the crash. The video helped me understand this particular crash.

 

[L_Mont]

In the video, the Tesla went from 77 mph to 55 mph in about 2 sec

 

[Silicon Desert]

The Tesla went from 77 mph to 55 mph in about 2 sec

Good info. I see the speed change. Not that significant. At any rate, that does not seem like too bad of a situation. I've seen worse hard braking from other cars. Yea, it is not something I would want to have. Those issues need to be remedied as we all know and hope.
Oh and thanks for the video. That was an interesting one.

 

[Silicon Desert]

I guess I will add a comment from the wife. She's processed over a few thousand auto accidents in her 30-year career. Her thought is that the 20-mph slowdown in not the "main" reason of the accident. If the car behind the Tesla had been traveling at a safe distance, they would not have rear ended the Tesla. They will be at fault in this instance, though people will disagree with that assessment. It is a rare situation where a trailing car is not at fault when they crash into another car. The missing piece of info here is not being able to see the distance the other car was training the Tesla.

 

[Tam]

...The missing piece of info here is not being able to see the distance the other car was training the Tesla.

When the Tesla slowed down from 77MPH to 55MPH, there were only 2 seconds left before the collision.

The rear view started at 00:04.35, as indicated on the video editor time counter in the lower center,

and the blue Acura collided with the Tesla at 00:04.96 or 61/100 seconds or slightly over half a second and under 1 second.

According to the Federal guidelines:

"Broken lines should consist of 3 m (10 ft) line segments and 9 m (30 ft) gaps"

And the blue Acura was 1 solid white line plus 1 gap away from the Tesla. That means 10 feet + 30 feet or 40 feet total away but at the speed of just under 1 second or 61/100 seconds or 61 milliseconds.

40feet/0.61 seconds x 3,600 seconds/hour = 236,065.57 feet /hour

There are 5,280 feet per mile.

236,065.57 feet /hour / 5,280 feet/mile = 44.70939 MPH

So it looks like the blue Acura did slow down before the collision.

On the other hand, the video editor might not be that accurate, and my math can be faulty too.

Let's review:

The Tesla purposefully slowed from 77 to 55 MPH in just 5 seconds.

Photo: The New York Times

The Tesla driver could blame the machine did it and not her. Although we call it phantom brakes, the system sensed an obstacle, which is why it drastically slowed down when it spotted the parking towing truck on the shoulder of the road.

Photo: The New York Times

Or from 70 to 55 MPH in just 1 second.

Photo: The New York Times

In addition to the purposeful machine's intention, the driver also legally agreed to drive a beta product knowing that it could get into an accident if the driver did not drive it. The driver did not press the accelerator to override the machine in this case due to confusion. Still, the driver intentionally used the beta product, so this accident was not accidental.

It's like when a driver takes in alcohol and can't react in time and causes an accident. The driver intentionally used something that could result in an accident, and it did, so that is no accident.

It's clear that the Acura is a victim of Tesla brake checking.

 

[dtdtdt]

I think you missed something: While the Acura was 1 stripe + one gap behind the Tesla at 00:04.35, both cars moved forward during the 0.61 seconds before the crash. So, the actual distance covered by the Acura during that 0.61 seconds was more (about double by my estimate).

Also, to me, it looks like the other car behind the Tesla in the right lane, was moving noticably slower than the Acura, suggesting that the Acura was booking it.

FWIW, it looks as if the Acura was not damaged much. If it had been just a little slower …

 

[Tam]

...both cars moved forward during the 0.61 seconds before the crash....

Ahhhh! Relativity at play! That does make sense. Since the Tesla was moving at 55 MPH, the Acura must overcome that speed to catch up.

 

[Supcom]

Am I the only one who noticed that the tow truck had flashing lights turned on? Perhaps the Tesla interpreted that as an emergency vehicle, which it is, and so moved back to the left lane and slowed to avoid it.

Nonetheless, the Acura was obviously tailgating. Generally, that's a slam dunk against the tailgater.

 

[nvx1977]

We def need to see more of the rear view. Another theory, based on the fact that you can see the left blinker flash shortly after impact: the blue acura was initially traveling at high speed in the middle lane. It saw the tesla transitioning from fast to middle lane, so not wanting to be slowed, the acura signaled and moved to the left lane. But then the Tesla moved back into the left lane and braked. Acura's speed prevents it from avoiding the crash.

Or, the blinker was a fluke (driver arm bumping the turn stalk immediately after impact).

 

[Tam]

Am I the only one who noticed that the tow truck had flashing lights turned on? Perhaps the Tesla interpreted that as an emergency vehicle, which it is, and so moved back to the left lane and slowed to avoid it.

Nonetheless, the Acura was obviously tailgating. Generally, that's a slam dunk against the tailgater.

Aborting a lane change is not the issue.

The issue here is baiting a traffic flow at 77 MPH then brake checking down to 55 MPH.

The Tesla could have aborting the lane change without disturbing the 77 MPH traffic flow.

It's the brake check that is the issue.

 

[Supcom]

Aborting a lane change is not the issue.

The issue here is baiting a traffic flow at 77 MPH then brake checking down to 55 MPH.

The Tesla could have aborting the lane change without disturbing the 77 MPH traffic flow.

It's the brake check that is the issue.

Nope. The tailgating is the real issue.

 

[hgmichna]

We can learn something from this. There is a particular type of accidents where a following driver extrapolates the movement of a car in front into the future, but fails to consider that the driver in front may change his mind.

A typical example is a car turning off into another street or driveway. Too many following drivers leave their speed so high that they can clear the turning car in front only if it continues without braking.

The Tesla Autopilot does not do that. It behaves such that there will be no collision even if the car in front suddenly stops. In other words, as long as even a small bit of the car in front is still in the lane, it considers the lane as blocked.

A good human driver will probably drive a little faster than the Tesla Autopilot in such situations, taking into account that there are still a few evasive maneuvers like steering to the side of the lane or even steering into the neighboring lane if that is free. The Tesla Autopilot would probably do something like that only if it has already got into an emergency situation, like if a car that turns off suddenly reverses.

The accident of this discussion is partly in this class. Had the following car been a Tesla on autopilot, the accident would never have happened, because the car in front had not completely left the lane.

Many Tesla drivers fault the autopilot for this behavior and consider it overly cautious. They forget that the autopilot, just as any human driver, can misjudge a situation due to the little time it has to do the right thing and due to all kinds of limited vision and limited mental (or computerized) processing power. Human drivers have phantom braking as well.

 

[Tam]

Nope. The tailgating is the real issue.

At 40 feet away?

 

[KLam]

The Acura was gaining the Tesla by 44 mph. i.e. the Acura did not slow down at all. And might have sped up instead.

 

[nvx1977]

At 40 feet away?

Not sure what you're even trying to say here. If you're following a car where you can't prevent a collision due to the type of braking the Tesla exhibited, you were following too close for the speed you were traveling. Tailgating is not some hardcoded following distance.

As is the law in the vast majority (if not all) of the states, the following vehicle is responsible for being able to avoid an accident due to situations in front. If you can't, you are at fault. Despite the flaws of the safety score, this is exactly why Unsafe Following was one of the factors, and it used time between cars, not distance, as time automatically takes into account speed. The faster you go, the more distance you need to keep to be able to stop in time.

 

[Supcom]

At 40 feet away?

At highway speeds, 44 ft behind another vehicle is severely tailgating. It's only 0.5 sec behind the lead vehicle. General guidance I learned many years ago was to keep a 2 second gap to the car ahead.

 

[Tam]

The Acura was gaining the Tesla by 44 mph. i.e. the Acura did not slow down at all. And might have sped up instead.

Thanks! Now that @dtdtdt mentioned. I think my method was wrong when I measured 40 feet from one single frozen picture because both cars were also moving so it's not just 40 feet but many more from many more frames of pictures too.

 

[Tam]

Not sure what you're even trying to say here. If you're following a car where you can't prevent a collision due to the type of braking the Tesla exhibited, you were following too close for the speed you were traveling. Tailgating is not some hardcoded following distance.

As is the law in the vast majority (if not all) of the states, the following vehicle is responsible for being able to avoid an accident due to situations in front. If you can't, you are at fault. Despite the flaws of the safety score, this is exactly why Unsafe Following was one of the factors, and it used time between cars, not distance, as time automatically takes into account speed. The faster you go, the more distance you need to keep to be able to stop in time.

At highway speeds, 44 ft behind another vehicle is severely tailgating. It's only 0.5 sec behind the lead vehicle. General guidance I learned many years ago was to keep a 2 second gap to the car ahead.

Initially, when I looked at the big gap away from the Acura, I thought that was a long gap. Now, I agree that one specific distance like 40 feet is not a standard to classify tailgating or not, it's the 2-second gap, so the distance is variable due to speed.

Even if the Acura was a tailgater, the law exonerates those who are victims of brake checking:

This article said slowing down even when you have a justifiable reason in the case of having a tailgater behind you, is illegal:

‘Brake checking’ a tailgater: Is it legal in California?

It’s a scenario commonly experienced on roadways: a driver from behind gets too close to one’s bumper, potentially resulting in feelings of anger or nervousness.

fox5sandiego.com

 

[Supcom]

This article said slowing down even when you have a justifiable reason in the case of having a tailgater behind you, is illegal

I have found that the best way to get rid of a tailgater is to slow down. Gently, of course, but if you keep lowering your speed, the tailgater eventually realizes it and will go around you. I use this technique often to mitigate PB risks.

It's curious how low your speed must get sometimes, as the tailgaters are so concentrated on your bumper that they don't realize how slow they are going!