I don't know what is stopping them from simulating everything inside a computer.
Record the input from all the sensors when a car equipped sensors is driven through real roads by a human driver. Replay the sensor input, with enough random variations and let the algorithms train on it.
Continue adding to the library of sensor data by making the sensor car by driving it through more and more real life roads and in real life situations. Keep feeding the ever increasing library of sensor data to the algorithm, safely inside a computer.
Obviously they've already tried that, and it doesn't work. The map is not the territory.
In theory, there is no difference between theory and practice. In practice, there is.
What I mean is that. Do not "teach" the thing in real time. Instead collect the sensor data from the cars a human is driving (and also collect the human input also), and train the thing on it, safely inside the lab.
You say, they have done it already. But I am asking if they have done it enough. And if that is so, how come the accidents such as these are possible, when the situation is pretty out of a text book in basic driving?
It also will not ensure exclusion of manual vehicles, so it won't create the exclusion necessary for the predictable driving environment.
Sounds like requiring exclusive access - I apologize if that was a misinterpretation.
If you have human and automated drivers in the same roads, the computers have to be able to cope with the vagaries of human drivers.
How can you then get away from '"beta testing" our self driving cars on roads with human operators' if that is their deployment environment?
I think we need to keep the human driver in control, but have the computers learning through that constant, immediate feedback.
And get rid of misleading marketing and fatal user experience design errors.