What will a driverless vehicle do when it arrives at a road section where, somehow, lanes aren’t marked? What if it arrives at a crossroad where the traffic lights are down and the police hasn’t arrived yet? What if there’s an accident on the road? What if there’s a policeman controlling the traffic and he’s not wearing white gloves? What if the movements he’s making or the way he’s standing aren’t according to regulation?
Artificial intelligence proponents tend to think that if you analyze further and further and finer and finer, eventually you will solve all such problems. At this point, artificial intelligence leaves the domain of computing and enters the domain of philosophy. One of the questions philosophy is concerned with is how humans find their way in the real world. Many, if not most, modern philosophers would disagree that we do so by learning to analyze finer and finer.
This is why I doubt that level 5 autonomous driving is possible under current and foreseeable technology.