[maxlin]

No? Public roads are fine. They've been fine for as long as they've existed. So the right option is to just make tech that works with them, and works like cars and roads have always been interacted with. By turning vision in to wheel inputs.

With what we can do with x amount of technology in given volume improving, increasing cost and making motoring just more unreliable (and to only work in cities) would be a strategy short sighted as running in a forest with your eyes closed. And sounds like Solar Roadways lol

[queenkjuul]

We already have to pave and paint and illuminate nearly every road. If humans have to take over on gravel that's not really a big deal, very few people drive on unmaintained roads

[lvl155]

Do you not see my point? How about traffic lights? Or street lights. It’s the same thing. You make it sound like things can’t improve from status quo. Let me guess you are a Java dev.

[maxlin]

As much as I hated boilerplate having to write some java last month for the first time in like 5 years, I hate other kinds of bloat and fragmentation as much.

I don't think we'd do away with street lights even if the cars could see in the dark. Traffic lights we'll need for as long as human driving isn't illegal, which is at least 50 years.

Safety is pretty much irrelevant unless its uniform with no large dips in intended design. Have existing roads. Have human driver cars, and autonomous cars which use all the same signals the humans use. These things don't advance as fast as some areas, so putting the cart before the horse has no benefit here. Even if we want to progress as fast as possible!

And there's streets outside cities too. The ugliness of fragmentation with your way is too painful to think of. Like soviet roads where suddenly its undrivable.

[trainsarebetter]

Good design is good design. Road dosnt need to be anything more than a path with consistent surface texture.

If we want to develop general purpose machines, then we can rely on placing external sensors in the environment