I don't want them to come back. One of the key advantages of buses is that they're flexible and could react to shock loads. Moving people in bulk should be done with grade-independent rail, like the Skytrain.
They're also much peppier, and have AC.
PDF Link to Map: https://www.valleymetro.org/sites/default/files/uploads/even...
Buses have to start/stop based on passengers desiring to exit, passengers waiting, and passengers completing loading. Drivers detect and react to those conditions, but it's not about whim.
[1] https://en.wikipedia.org/wiki/First_Hill_Streetcar#Vehicles [2] https://www.king5.com/article/news/local/sdot-pulls-first-hi... [3]https://www.seattletimes.com/seattle-news/transportation/sea...
For buses, you might be interested in this: https://www.oppcharge.org/
I guess detailed calculation using future projection is necessary to make a case for either one. The choice would also depend on local circumstances.
Buses are probably an 'easier' problem, but it's much easier to pitch selling self-driving cars to individual consumers rather than to government procurement budgets. Also, I suspect the bus problem space introduces additional issues such as <what to do when some asshole is parked in the bus bay>.
It'd also be nice if more of the self driving companies would go after busses first. Fixed routes seems like a significantly easier problem and you can amortise all your sensor/electronic cost over a more expensive vehicle.
Both combined would pretty radically alter the environmental and economic cost of public transportation.