Tesla takes out Google autonomous vehicle

We should mention that the report indicates the Google vehicle was being manually driven when it was hit. That said, as to why in general Google AVs are rear-ended, it is because they actually drive safe and decide not to run over pedestrians or bicycles (the former being what happened this time).

Note the accident took place last August 20th 2015 (before following the link I thought this was something that just happened).

Also, I feel for the Model S driver, from the description they were changing lanes (and, likely, taking a quick shoulder-check at what turned out to be exactly the wrong time), and I've certainly heard about / been in close calls before in the same situation where 1/8th second will make all the difference.

Ugliest1 said:

Also, I feel for the Model S driver, from the description they were changing lanes (and, likely, taking a quick shoulder-check at what turned out to be exactly the wrong time), and I've certainly heard about / been in close calls before in the same situation where 1/8th second will make all the difference.

Click to expand...

This seems likely and personally I would give them the benefit of the doubt and assume this (especially given no report on whether the light was green or red) unless otherwise told so.

electrish said:

The big question though is, why do Google's autonomous cars constantly get rear-ended?
Report of Traffic Accident Involving an Autonomous Vehicle (OL 316)

Click to expand...

The stories this week have all been about how Google was looking to make their autonomous cars drive more like people. The apparent issue is that they've been overly conservative and cautious in questionable situations. If an autonomous car stops unexpectedly out of an abundance of caution in a situation where a driver might only have slowed or turned around an obstacle, you can understand why the incidence of rear-end collisions would be higher.

99.9% of rear end collisions occur because the following vehicle has left insufficient room between their vehicle and the one in front of them. It is that simple. Judging by comments made on another thread relating to where people set the safety gap on their TACC, you really don't need any more proof that this is the case. If you do, just go and watch the freeway traffic for a few minutes from an overhead bridge or similar. MW

martinwinlow said:

99.9% of rear end collisions occur because the following vehicle has left insufficient room between their vehicle and the one in front of them. It is that simple. Judging by comments made on another thread relating to where people set the safety gap on their TACC, you really don't need any more proof that this is the case. If you do, just go and watch the freeway traffic for a few minutes from an overhead bridge or similar. MW

Click to expand...

By definition this is true. In theory I have to be able to stop in time even if the vehicle I'm following stops in 0.1s otherwise I'm at fault. The reality is that vehicles don't come to a stop that quickly and it's usually inattentiveness or miscalculations that lead to the accident rather than braking distance issues. That, after all, is what safe following distances are for ... to account for human reaction times and ensure that slight errors don't result in collisions. However, attentive drivers don't really require 7-10 car lengths to react to a change in traffic flow.

JohnQ said:

However, attentive drivers don't really require 7-10 car lengths to react to a change in traffic flow.

Click to expand...

Following distance is to allow for the driver reaction time plus car brake stopping distance, and distance covered is related to the traveling speed. Therefore 7-10 car length is not a definitive following distance.

martinwinlow said:

99.9% of rear end collisions occur because the following vehicle has left insufficient room between their vehicle and the one in front of them. It is that simple. Judging by comments made on another thread relating to where people set the safety gap on their TACC, you really don't need any more proof that this is the case. If you do, just go and watch the freeway traffic for a few minutes from an overhead bridge or similar. MW

Click to expand...

Lets also be clear on the topic of driving characteristics in the US vs. Europe (I see you are in the UK). In short North America is atrocious compared to most other countries in the EU.

Drivers clog up as many lanes in parallel with zero consequence or understanding of the impact that it has. you will see 5 lane highways in California that aren't busy but they are clogged with 10 cars in a pack blocking all lanes and going roughly the same speed - not to accelerate unless presented with the chance that they might be passed. They pass on the right and just as dangerous are the speed regulators that jump over to the "fast" lane and insist in clogging it up going as fast and sometimes slightly slower than the car to their right or the posted rate of speed.

Tail gating becomes a natural by-product and dangerous chances are taken. It clearly wasn't the case in this accident report, but the 99.9% factor you refer too has SO much bad behavior behind it, Im not convinced its as simply as driving too close as the only factor.

North America needs some revamping in general about what is acceptable driving.

- - - Updated - - -

Soolim said:

Following distance is to allow for the driver reaction time plus car brake stopping distance, and distance covered is related to the traveling speed. Therefore 7-10 car length is not a definitive following distance.

Click to expand...

7 to 10 car lengths leaves room for 10 cars behind you jam in front of you while you're busy being safe....if you choose to do this, don't do it on the far left lane of a freeway. Otherwise known as the "fast" lane

Soolim said:

Following distance is to allow for the driver reaction time plus car brake stopping distance, and distance covered is related to the traveling speed. Therefore 7-10 car length is not a definitive following distance.

Click to expand...

Yes, distance covered is related to speed and I'm exaggerating a bit with 7-10. But my vehicle stops at roughly the same rate as the one I'm following meaning my reaction time is the primary factor (plus the difference in speed under full braking as a result of that delayed reaction).

My point about all of that, though, is that we have a built in sense of predictability ... for example, slam on the brakes hard initially in a panic situation then modulate the braking based on visual feedback. Or slowing to a speed that will allow the car in front of me to compete a turn into a driveway while not unduly delaying the traffic behind me. That latter one is particularly tough because you have to judge the speed of that turning car, whether there are any other obstacles that will prevent its turn, etc. Another is the natural tendency I have to slow down and focus when a child pedestrian approaches a curb since they are more unpredictable than an adult approaching a curb. Autonomous vehicles have a great deal of learning to do to manage the many scenarios and judgements that experienced drivers do as a matter of course.

There are othter factors. I have been rear ended twice. First time the driver was under the influence. Second time the driver's vehicle was not in good condition (worn tires and possibly bad brakes ).

Not in Model S.

This post made me laugh as a quintessential California event!
Where else in the Universe would an autonomous Google vehicle rear-end a Model S?
When I first moved to California, my wife and I were both rear-ended in the first month (2 separate incidents). Never rear-ended before and never since (so far).

DFibRL8R said:

Where else in the Universe would an autonomous Google vehicle rear-end a Model S?

Click to expand...

The Google car was rear-ended, it didn't do the rear-ending.

JohnQ said:

Yes, distance covered is related to speed and I'm exaggerating a bit with 7-10. But my vehicle stops at roughly the same rate as the one I'm following meaning my reaction time is the primary factor (plus the difference in speed under full braking as a result of that delayed reaction).

My point about all of that, though, is that we have a built in sense of predictability ... for example, slam on the brakes hard initially in a panic situation then modulate the braking based on visual feedback. Or slowing to a speed that will allow the car in front of me to compete a turn into a driveway while not unduly delaying the traffic behind me. That latter one is particularly tough because you have to judge the speed of that turning car, whether there are any other obstacles that will prevent its turn, etc. Another is the natural tendency I have to slow down and focus when a child pedestrian approaches a curb since they are more unpredictable than an adult approaching a curb. Autonomous vehicles have a great deal of learning to do to manage the many scenarios and judgements that experienced drivers do as a matter of course.

Click to expand...

The guideline I remember is not really 7 to 10 car lengths. It is one car length (or more) per 10 mph of speed, where the "more" is needed in case of poor weather/visibility.

Max* said:

The Google car was rear-ended, it didn't do the rear-ending.

Click to expand...

Either way, only in California

DFibRL8R said:

Either way, only in California

Click to expand...

What this little accident highlights is that California has supplanted Detroit as the American center of leading edge automobile technology. Of course some of the major European manufacturers are also at the forefront of technological innovation (along with Mobileye in Israel, the Japanese not so much) but where else in the world besides California have dozens of autonomous cars been on public roads for so many years?

The European guideline is to have 2 seconds length gap to the car in front of you. Pretty simple.