I can’t help thinking that the only way it will really work is all or nothing - if driverless and driven / other conventional road users (pedestrians, cylists, motorbikes etc) are segregated - some roads or lanes for driverless only, for example. Otherwise, there will always be the potential for a mistake or some deliberate action by a clever cunt to screw up the driverless system.
Think of the horse on a country road, wobbly cyclist who hits a pothole when the car is 5m behind them, the arse in a hurry who pulls out of a side road / lane on M-way into a gap that’s too small, knowing that the driverless car that’s approaching will slow down for them, or whatever takes your fancy. Will the programming predict the combinations of events that can lead to dangerous situations and act ahead of time to reduce the risk, as we can? Likewise, will it be polite when another road user needs a little help to get out onto a busy road, past an obstruction etc?
The plod, ROSPA, IAM and others train people to think ahead in this way, but I fear it will be some time before sensors, programming and vehicle networking will give the same sort of safety performance that we could have already. [Brief jaunt on hobbyhorse; why the hell don’t we all do a re-test every 5 or 10 years, or after a significant event / accident? Why aren’t we expected to upgrade our skills to take account of changing technology, traffic and road conditions? Why is there so little incentive for us to do appropriate advanced training to prove that we can be safe drivers?]
I’m watching with interest, but you’d have to pay me quite a lot to travel in a driverless vehicle at this point. I’d want to know a lot more about the design safety requirements, test, evaluation and acceptance procedures by the manufacturers and real-life safety performance before I go near one (must watch the Guy Martin thing)