If there aren't enough automated cars in a fleet to get everyone to work bang on at peak hour time, then you can't hire an automated car to get you to work. This is the frequently cited problem with car sharing, what happens at peak times. You have to buy a private car, incurring a heavy capital cost ($50k, in this example). Most people don't actually have to travel at peak hour though, there is nothing stopping you from commuting to work earlier and avoiding peak times. Now you don't need to buy a car. That is the example.
You did imply that total maintenance costs of a car are around as much as capital outlay, by suggesting there are no significant cost savings to be had by sharing a car. Explain how, if sharing a car is not cost effective (because the increased maintenance offsets any saving), how it could be interpreted any other way?
In cases where technology has reduced risk, insurance prices have come down. There is no other way it could be in a healthy free market, otherwise I would open up an insurance shop and cash in, collecting high premiums and rarely having to pay out. How do you think insurance premiums are calculated? Someone just pulls a number from thin air, that sounds good for profit margins? No, market forces dictate what the premium will be, based on how much the insurer would be expected to pay out.
Rego, and taxi medallions, are free market distortions, but again this cost is reduced by sharing it with others. These costs will go up, as governments claw back lost revenue from fuel excise, but this is nothing specific to automated cars.
