That’s true on paper, because computers can determine distance and speed of the car in front much faster than human processing time, and can also react to changes much faster than a human. But in practice I’m not sure it’s implemented on the boundary. We just bought a car that automatically brakes if it thinks you are getting too close to the car in front. It consistently leaves a much bigger margin than I do when I drive. I’m not talking about blatantly unsafe practices like tailgating–it brakes sooner than I would in situations like normal stopping when I am queueing at a red light or the car in front is braking. In many decades of driving I have never rear-ended another car.
(bolding mine)
I think that is economically not feasible (think of the delta in riding fees w/ vs. w/out pilot!) … if you have to employ a highly paid professional pilot … (as opposed to a “free” AI-pilot)
I presume in congested areas would have different levels of flight and there would be approved “corridors” for each direction, in lieu of traffic lights; and approved descent vectors that the corridors bypass, to specific landing spots. Perhaps flying cars would be sufficiently expensive that congestion is less of a problem. But a mess of random point to point flights seems frightening.
I prefer the Musk vision of a maze of one-way tunnels criss-crossing under the streets, where only self-driving cars are allowed zipping along at highway speeds under AI traffic control; if necessary, ramp up to the surface roads to complete a journey to less busy areas.
There is supposed to be a macroscopic relation, valid over a certain-sized region, relating the traffic flow to the traffic density. E.g.
though the actual shape of the curve will depend on various things.
Theoretically, will more computer-controlled cars indeed skew the curve to the right?
Could you explain that graph, please? All I can tell from it is that there’s something we’re trying to optimize, and that it has a maximum somewhere in parameter space.