Uber should release its camera footage of its fatal crash to the public

wordpress ^ | March 20, 2018 | Dan from Squirrel Hill

[grundle]

Uber should release its camera footage of its fatal crash to the public

Two days ago, a self driving Uber car crashed into and killed a woman in Arizona. The car had a backup human driver behind the wheel who had the ability to take control at any time. The woman who got killed was walking in the street but was not in a crosswalk.

In my opinion, the government should get a warrant from a judge to require Uber to release its camera footage of the collision to the public. As long as we don’t get to see the footage, we can only speculate as to who was at fault.

If it was in fact Uber’s fault, then the public has a right to know, and Uber should be required to pay $10 million to the family of the victim. (I also believe that anyone who fakes such an accident in order to commit insurance fraud should get 10 years in jail for insurance fraud, in addition to whatever punishment they get for killing someone.)

If it’s the pedestrian’s fault, then knowing this information would prevent people form mistakenly thinking that self driving cars are more dangerous than they actually are.

So far, Uber’s self driving cars have a death rate of one death per approximately 2 million miles. By comparison, human driven cars have one death for approximately every 100 million miles. These are just rough numbers – they are not exact. And the sample size for Uber’s self driving cars is too small. However, from what we know, so far, Uber’s self driving cars have a death rate per mile which is approximately 50 times that of human driven cars. If this death was the fault of the pedestrian, then it doesn’t give any reason to be afraid of self driving cars. But if the death is Uber’s fault, then it’s a sign that something is seriously wrong with Uber’s self driving cars, even though the sample size is small. In cases of life and death, even one death is too many when only 2 million miles have been driven. The sample size is small, but that doesn’t change the fact that a person is dead.

 

[Ozark Tom]

Given an average human reaction time of .25 seconds for a visual stimulus, an automated system should respond at least one hundred times faster. At 35mph that removes 13 feet elapsed for driver reaction time. With an average car this yields a typical 61 foot stopping distance for the A.I. The sensors should be able to detect a hazard at a distance commensurate to driving speed to allow for a successful avoidance of a collision.

Does the A.I. brake only, swerve if unencumbered by other traffic, or do both? Can it predict that a human cannot quickly reverse course, then evade a collision by swerving behind the person’s path of travel? Are they forbidden from entering an empty sidewalk in an avoidance maneuver?

Continuously evaluate the path ahead. Proceed only if the senses match the model. If in doubt—stop. Initiating a stop will add more time to perceive the actual condition and initiate corrective action.

https://www.random-science-tools.com/physics/stopping-distance.htm