Some observations, since I’m presently designing in a domain very close to the one being discussed: autonomous vehicles are on the way whether you like it or not because of socio /economic / technological imperatives. This survey is more to throw some of the associated issues into the public domain because legality and morality will both have to be modified to accommodate this new thing.
Stop thinking of the car as an individual. In a peloton of autonomous vehicles your car won’t be. It will be making decisions as part of a network at a rate so fast the actions will be enacted before the occupants have any clue something is happening. Imagine a mini train formed of autonomous vehicles tailgating each other at 100 mph on the highway, all braking and manoeuvring simultaneously while all the occupants in all the cars are having beer or coffee and talking (or more likely, busy with their smartphones). It’s about as far from traditional road transport as you can imagine. Forget everything you’ve learned from that.
the only things that are really needed is hardware in the road ie, markers on the lanes, and intersection signals that can be read, the way tesla is doing it now is just the beginning.
This is actually the wrong war around. The earliest autonomous vehicles took this approach. Citroen had a functioning autonomous DS in the early 1960s which applied this philosophy but the cost of modifying the infrastructure is too high. The autonomous ‘thinking’ and sensing will mostly be onboard with of course some external digital knowledge of traffic conditions, sensor-pooling etc.
I have seen many, many problems caused because of the hubris of engineers who believe their designs are infallible
Interesting observation. Like to discuss your experiences sometime. Engineers ought not be doing the design work in some domains as they presently do. In the building industry for example the tasks are more formally delimited. Of course good engineering should also allow for failures in a failsafe condition, where possible. Accidents will happen regardless.
Regarding the discussion of who should cop it in an emerging accident: presently road accident-related casualties considerably outnumber those caused by wars and murders on a yearly basis. Given the ability / attention / interest in driving of an average human operator, even conceding the fact that humans can use intuition and interpolation AND there will be mixed autonomous / non-autonomous vehicles flowing together do you think the accident rate will go up or down?