I generally agree that self-driving cars are going to take this niche, but not with tunnels. Tunnels add the same dedicated infrastructure problems as mass public transit.

I'd suspect most car trips today are 1 or 2 passengers with the back seat and trunk empty; we'll eventually see new form factors of on-demand vehicle that trim off unneeded space. If you need to get from A to B alone, no cargo to speak of, you order a ride that covers that class and it's small. If you're taking a shuttle from the airport with your whole family and luggage, you order a ride with those specs.

If you are not being facetious, what you are describing is closer to a subway system, which has the disadvantage of being very expensive.

Hopefully someone else, so it actually happens and isn't overpromised and underdelivered.

(Also, tunnels are useful not just for the increase in traffic, but for moving car traffic away from non-car traffic, which makes both kinds of traffic safer, faster, and more efficient.)

As long as one of those points is a transit stop then yeah, robotaxis make sense. In that model you don’t need the tunnels.

They make even more sense if they are a bit larger and can accommodate multiple people at once. Something like a large van or small bus.