Theirs lots of videos of FSD drivers going round San Francisco with streets comparable to what you get in London, you forget it's a fairly dense and pedestrian and cycle focused city, not endless suburbia. Its currently only available in places that tesla have mapped out and approved to run, not expected to be allowed everywhere.
My brother regularly sends me videos of him been driven by an entirely self driving Waymo taxi (that doesn't even have a support person sitting anywhere in the car.) He lives in Los Angeles, California. So it seems these vehicles have advanced to a level where they can work independently.
Theirs not entirely zero human caused car crashes. If you can prove self driving is safer than human driving, but still keep them banned until they are perfect, you are accepting extra people dieing for the sake of luddism
Numerous European metro systems have been self driving for a fair ammoumt of time, hardly a tedious novelty. London has had a fully self driving railway since the 80s and a number of tube lines are basically self driving.
Considering how popular Waymo is , I highly doubt your assertion is true
The Vehicle to Everything (V2X) protocol is a mesh system, so communication is done in real time, from car to whatever is chatting back. Volvo have begun fitting it in Denmark and is happily warning other cars around it if a car finds some slippery road surface. Mercedes are fitting it to E and S Class and have a good write up here. They describe temporary traffic lights that can talk to your car to let you know when the red light will turn to green. And advance warning of an ambulance coming from around a corner. And the mesh network of cars randomly driving around a city that are also scanning empty road side parking spaces and reporting to anyone looking for a space that there is a space 2 roads away. I'm sure there are hundreds of other uses.
I’ve swerved up a verge and through a bush to avoid the back of a van once, and another time someone pulled out toward me and I had to swerve round into a kerb (which broke the suspension and left me careering into a grass bank). It doesn’t have to happen a lot - it’s the precedent set by one big court case and who they decide has to pay up. Lawyers are going to make a lot of money setting up the principles and then authorising the choices manufacturers/programmers make.
As i previously mentioned I think there is far less appetite for people being killed by robot’s choices than ‘human error’.
Look on the bright side. If these automated cars are so great, we can stop spending money upgrading level crossings because they will never try and jump the lights or barriers.
That sounds great on paper, but if you think about it for a moment the implications are worrying. Temporary traffic lights are set up by roadworks crews countless times a day across the country, and are fairly simple. If there's a load of extra kit to allow the lights to communicate with autonomous vehicles, who will provide this kit? Who will train the crews on setting it up and who will pay for the training? How will these installations be monitored to ensure the lights are being placed and calibrated correctly, and that those installing them are qualified to do so?
Why would their be much extra work for the crews? It would all be handled by the computer that controls it that requires minimal input to set up. What would calibration mean in this context, it would be a digital radio signalling protocol not an old television set
I was cycling home one day, and as I approached a temporary light, it turned from green to red, then back to green very quickly after as I approached. I suspect it set both directions to red then allowed someone through when detected by the sensor. If programmed correctly at the factory I don't see how this is much more difficult to set up than a "dumb" light, given that the sensor was integrated into the traffic light face. It would be nice to see permanent lights this advanced.