[misterbwong]

I've asked this around a few times and I don't if it's completely dumb or somewhat viable.

Why can't we put rails on top of (some parts of) freeways and adapt our cars to work somewhat like railways? Call it "smart freeways" to sell it to general public.

When you merge on to the freeway, you "hook in", and the car starts to control itself. You tell it where to exit, and it takes care of merges, changes lanes, exiting, rerouting due to unforeseen obstacles, etc.

Seems like it could end up with the benefits of self driving cars but with training wheels and utilizing existing tech. Cars could easily signal lane changes, merges, and generally communicate with other cars to make things flow evenly and allow vehicles to move around. You would also avoid the "last mile" issue of getting you from the freeway to the destination-that would still be up to the driver

[kayodelycaon]

Actual metal rails aren't viable for a large number of factors. The least of which is maintenance due to big trucks damaging continuously degrading road around them. You'd be better off putting magnets in the road and have equipment in cars to follow them.

Even then, you're no better off than existing self-driving because you have to deal with every other car on the road that doesn't support "rails".

[misterbwong]

Ah, yes I forgot to mention that part of the idea is that everyone on that particular "smart freeway" would be required to have the hardware to support it, no exceptions.

Yes, I understand that would be quite regressive and has a host of other concerns. However, for the narrow purposes of easier, safer, and more efficient driving while on the freeway, I do still wonder if it will work.

[kayodelycaon]

Assuming your suggestion is possible, why wouldn't we just replace the road with railway tracks and move vehicles on flatbed cars instead?

No one would need to change their cars.

And before anyone mentions throughput, let's assume we can use custom track sizes and just have people drive into the side of a railway car, like it was a ferry. Should be able to stack cars two height.

[misterbwong]

I mean...you're not wrong on that suggestion :D It just seems like _something_ would be easier than solving for every.single.edge.case with generalized autonomous driving.

[withinboredom]

We can’t even get trains to drive themselves.