If you could get that robot-cooks-as-you-drive idea right, even without the self-driving bit, you can out-compete Dominos by saving half the trips (back to store from delivery address).
The technology was immature but the fundamentals were there, and they'd save a not-insignificant amount of gas.
That's easy. Put it on treads, and add a tree cutter/stripper, and a hopper, so that it can pick up all the fuel it needs to wood-fire the pizza off the side of the road before it goes to the next destination.
I really can't imagine that. So you save some driver time. They're usually idle before delivery already, there's no driver shortage, and drivers/delivery isn't expensive. So there's only a small amount to undercut. For the profit to become interesting, you'd need a massive volume, but then you'd be forced to maintain a large fleet of mobile ovens. There's no automated driving, so you'd need drivers anyway. You'd have to pay everything from their idle return time, including the increased energy consumption.
The fact that Dominos practically begs people to come in and pick up their pizzas themselves undercuts your arguments about driver idle time. Maybe in your market they sit idle, but in mine I'd imagine they are spending 90%+ of their time fulfilling orders. My orders often spend 10min in the "Quality Check" state (waiting for a driver under a heat lamp).
Automated driving is orthogonal to the bake-as-you-drive model. Dominos will also switch to self-driving when they can.
The cost of owning 10 oven-vans vs 1 store + 10 regular cars will be the tough part and will require scale, but pizza is big business.
Arch that "quality check" thing bugs me to no end. It's the lie that bothers me: it's not like they're individually inspecting every single pepperoni or anything. They just don't have a driver available. Don't lie to me!
The technology was immature but the fundamentals were there, and they'd save a not-insignificant amount of gas.