Not confusing. It is so much easier to make electric vehicles self-driving compared to the ICE vehicles. It may be economically feasible to do so in not so distant future. Real estate, including parking spaces is expensive. There was a good Freakonomics episode about that few years ago [1].
Main barriers to self-driving will soon be juridical, not technological. This is not just random internet rambling. I'm an engineer and have been working on commercial autonomous vehicle subsystems for a while. I can with full confidence say that the technology is already there.
Normal acceleration and maintaining speed is trivial to handle with simple control loops. This isn't like e.g. phase detect autofocus where you want to supply a single input at the start to get to the correct end position; accelerating a massive vehicle takes a significant amount of time, such that a control loop has to:
- gradually increase the accerator input from the start point
- decrease accelerator input as the desired speed is approached
- maintain a constant speed
This is trivial for a control engineer.
In some ways ICEs are easier than electric motors - you get free damping on your accelerator input!
Outside of normal driving, you also have emergencies, where surely the alternatives are basically braking, steering, and hard acceleration, which respectively require engine inputs of 0 (i.e. no gas), no change (or again 0), and max. And hard acceleration is almost never the answer to emergencies anyway.
When does a self driving car need to know "provide exactly X power to the engine", where X is none of [0, a little less than currently, no change, a little more than currently, max]?
Source for control loops: On a general engineering course bachelors, our 1 days labs included one of which a part was writing speed control software for ICEs and one of which a part was writing position control software for an electric helicopter.
