Doesn’t Lyft work sorta like that?
I've only ever heard of lyft being a normal taxi service where people just use their own cars they already own. Also, I dunno where you're getting your numbers for the calculation you're doing, that would probably be something good to include. You could say the same for everything I write, I guess, but none of my criticisms much have to do with the numbers, except for this: I dunno what "smaller european cars" you're using. Most cars nowadays are like, 2 tons or so at the least, probably more, and you could maybe get one ton of human body weight, at the most, if you had several 250 pound chucks riding around in one car, which I don't really imagine to be the case normally at all.
There's also an efficiency created by the "inefficient" route planning of the bus. By having something that travels in a loop, rather than having every individual travel to every individual point, we're trading some amount of efficiency in terms of total time spent by everyone (theoretically, but this time is probably eaten up by increased amounts of car traffic in reality), and we're trading that for a slight increase in the amount of foot traffic that people are collectively engaging in, which is probably a good thing. So that's a total decrease in curb weight as a factor of total travel time, which is a decrease in road maintenance.
You're also probably looking at a massive decrease in mechanical maintenance for buses compared to cars, using one big engine, set of brakes, A/C systems, etc, rather than like 15-20 smaller non-standardized sets, and maintenance costs for the specific roads you're traveling on via bus means you can engineer in less maintenance over time compared to a more spread out system.
Density is also a pretty big consideration, because real estate downtown, i.e. the location most people are going to want to go, is at a high premium, both for people and for the city/state's tax base. High density has the capacity to provide a sustainable tax base for the cost of providing utilities and maintenance by the city.. Unless you park the series of autonomous cars all in some huge superstructure outside of town, and then basically just merge them straight into the highway, where you'd still have to overbuild and deal with a massive amount of car infrastructure (more than just the space you'd save on all this parking, since you could just have a couple pickup and dropoff spaces, if that, compared to all this other parking taken up downtown). I can't really see it working out, and even at the normal densities we'd be looking at, I'd struggle to come up with a way by which it's more efficient overall.
There's also other types of buses, if we're just talking about emissions efficiency, or energy efficiency. Obviously an overhead electrified bus is probably the most desirable, just behind a tram or a streetcar or whatever. Then you have the weird stupid hybrid battery overhead-electrified buses that I hate, and then probably all your natural gas buses and diesel buses and whatnot, and then your pure battery buses.
If we're talking about autonomous vehicles, then we're kind of also sidestepping all these questions about like, the scalability of the AI for this, and the computing power we'd have to use on that, constantly. We'd have to deal with the traveling mailman problem on a near constant basis, something which public transport can mostly sidestep by assuming passengers will come to it, and that public transit will be of a high enough density to create desirable locations simply by stopping there. We have all the pedestrian and cyclist traffic conflicts which we'd encounter, or else have to segregate from these cars entirely (something normal traffic already struggles to do adequately). And if we're segregating the traffic entirely with a large amount of infrastructure, which definitely makes this much more achievable and easier, if still not easy, I think it makes more sense from a top down maintenance perspective to just go for trams or streetcars, or subways, or something like that.
I think the only real way in which I can cook up a reason this might be done, is because it's outsourcing costs onto the public, and onto the state. Road maintenance can be done by the city, or state. Probably, this would mean that the autonomous vehicles would not be segregated, which means it's less of a good idea, which I believe, is the primary reason it hasn't been done. Then, the taxi service could basically make a bunch of money on their highly necessary transportation, which they have created a large need for, simply by existing and demanding a large amount of infrastructure by existing.
Use bicycles, e-bikes, and walking for individual pedestrian point to point travel. Fuck all the bullshit excuses people give about how, oh no it's too hot out, too rainy, too hilly, what do I do with this cargo that's not large or consistently arriving or departing enough to be loaded by a freight train, or by a professional truck, but isn't so small that I can carry it, what do I do with all my kids, etc.. Use cars sparingly enough to fill the very minor amount of gaps that can't be bridged by bikes, cycling, and public transit, as a method of last resort. Mostly for people that would maybe need to live out in the boonies, like park rangers, maybe. Actual farms, not the stupid rich people playtime "ranches", and industrial locations, they usually have a large enough cargo haul to justify a small freight train, or a large truck taking a small route to a freight yard.
Why not just use a big shotgun for the smaller stuff, skeet shooting style? I dunno how to really deal with much larger drones, but a shotgun should be able to deal with most human sized drone targets pretty easily