“It means that the ‘self driving’ car has an incredibly limited ability to process data and make decisions ‘on-board’, and it relies on pre-mapping and pre-planning by a team of people, much like a space mission.”
The “incredibly limited” data-processing ability is (at least) doubling every two years. It’s already millions of times more powerful, than the on-board Apollo mission computers.
Google Maps, every GPS receiver, etc. already collect data, in real time, from their millions of users. That data goes to update road conditions in real time. Similarly, every autonomous vehicle will collect new data on road conditions, road markers, and other characteristics of the environment. The information will be available to all other autonomous vehicles.
Raise your hand if you’ve ever worked with LIDAR.
Oh wow, my hand is raised. So I’ve got a notion of how large the data files from a LIDAR scan can be. I also now that the one venture my company is working on that involves on board processing (not in an automobile and I really can’t say what) includes an air conditioner that would normally be found on a reefer trailer, in order to cool the stack of computers.
Again, without the advantage of pre-mapping a route, a Google car could not make it across a parking lot on its own. This is where we are at now - zero point. This is why I find the fantastical claims (and snarky comments about non-believers) to be so interesting.