Self-parking cars came before self-driving cars. What do you see is holding this back? Is it the lack of accurate maps? Is it the understanding unusual road signs or detours? Is it the ability to respond to unpredictable road conditions and obstructions? Is it all of the above? I take it you're not a believer in an imminent technological singularity. Neural networks have shown some remarkable abilities to learn new information without having to be specifically programmed for every outcome. Couldn't they be applied to the problems above to solve them the same way humans solve them? Do you think that will take many decades? I think if we even get close, they will have a huge advantage in that, as soon as one car learns, every car can learn the same thing because it's easier to copy and transmit technology than biology.