The biggest impediment is pretty simple, and its a combination of economics and legal issues. The gap between a car that is 99% autonomous and a car that is 100% autonomous is enormous, but that last 1% is critical to realizing the practical benefits of the technology and to resolve the legal issues associated with it.
If a car is not 100% autonomous then it will never replace the driver entirely, and if the driver isn't removed from the operation entirely then the technology makes no economic sense for most potential customers. You might be able to convince someone to pay $50,000 for a fully autonomous car instead of $30,000 for a "regular" car. But if the $50,000 car is NOT fully autonomous and still has to be operated by someone who must sit at the wheel and pay attention to the road in front of him, then what exactly is the extra $20,000 getting for the driver except a whole bunch of sensors and warning indicators that tell him to do things that he then has to do himself anyway?
You might be right about the “1%” impediment, but...
There has NEVER been any technology developed that provides 100% certainty of working as advertised or as desired. The 99% that works as desired or as advertised, would still provide huge improvements over current driving, and would provide for a lot less expensive “cost of living”. Heck, people spend a large part of their lives worrying about their personal transportation, and if we removed those worries, people could/would find a lot of other endeavors to spend their lives doing.
So, tell me, how safe are people-driven vehicles now? The biggest problem with current vehicles is not the vehicles themselves, it’s the people behind the wheels of those vehicles. I would say that, the “error rate” for people-driven vehicles is many times the 1% that you mentioned.