This technology WILL happen and it will have a better safety record than humans.
The driver herself would have hit this woman who only appears in the headlights at the last second. A computer radar can and should have seen her even before that.
It might even have detected her- but the only action to miss her would have involved a major swerve of the car, possibly to the point of tipping over.
Imagine if this was a deer jumping in front of the car- does the computer swerve to miss a deer and potentially dangerously impact the car itself or cars around?
I am NOT comparing a deer to a human, I am just explaining this in another light.
If ANY OBJECT suddenly swerves in front of the car, what does the human OR computer do? Suddenly swerve?
The computer driven car hit a WALL not a person.
The answer of course is that for the computer there are eons of time to make a decision. A thoughtful one that takes into account all the actions and situations. (Recall the scene in Terminator 1 — when someone knocks on the door and he refers to a drop down menu of possible responses.)
It is an important question as to what the computer should do. (I once decided to kill a raccoon rather than swerve on a freeway at night.) I never regretted the choice, and I basically had planned on this type of choice while practicing “defensive driving”. The car would be programmed to do the same. (This one was not that far along).
The point is that the car should have sensors that beat the human sight by a good margin,. (And I believe we will find out that it did but was not programmed with effective decisions. (Did the software go into a loop?)
We have a long way to go but the computer will get there.