My hunch is that ultimately, ceding control of the car to an automated system is going to produce better results, at least when driving on major "pipeline" roads. The biggest source of traffic jams isn't necessarily the pure density of cars, but the inefficiency of the individual driver. If each person takes that extra second to accelerate, that gums up the whole works (in NYC we are trained to be more efficient - taxis will honk at us if we take more than .05 seconds ;)
Imagine this: You enter a major arterial roadway -- I-95 or Route 80 and you program your destination -- an exit #, a rest area, etc. Your car joins the pipeline, syncs up with all the other cars on the road, accelerates to a (fast) cruising speed (100+ mph), and you sit back and enjoy the ride. No lane-changing, no "human" failures (like slowing down just because the sun changes direction) - and when it comes time to arrive at your waypoint, the system safely gets you out of the flow and into a "manual" lane. Then you take over from there.
Interestingly, this seems to converge towards traditional ways of mass transit, like trains. Except for two crucial differences with regards to the US: Insane connectivity, seeing how good the highway system is in the US, and of course the fact that once you get off the freeway, you still have a car to navigate through the vast suburbia (even if you have to do it manually). I wonder if this will be the replacement for good mass transit in the US, at least for now.
I think the final piece of this puzzle is going to have electrified highways so you can travel long distances without the need for large battery packs. It would still be more efficient to uses busses/trains but we already have a lot of roads and cars which make retrofitting seem like the best option.
When we got a talk from a renewable energy / car battery researcher at my university at the end of the talk I asked him "don't you think that given the costs and difficulties with scarce materials that it would be cheaper and more energy efficient to create power lines for cars on every road than to put a battery in every car?" -- and he agreed, except that this strategy doesn't have a good business model unless governments take major steps.
One of the other benefits is that a car controlled by a computer should be able to draft off the car in front it. This should significantly improve gas mileage. We might even design the shape of cars specifically for this since aerodynamics don't matter as much at slow speeds.
In the era of rising energy costs, there's still the fact that you have a one-to-three ton steel cage just to transport a 100-300 pound person. There's no way that's a sustainable solution.
I agree. The only problem is that you still need to leave room in case something catastrophic happens in front of you (deer, tire flat, etc.) Of course, if the system detects these events as slowing down of one vehicle, it could quickly slow others down, so you'd need less distance than with a human driver. But it only works so far. There are laws of physics dictating the distance needed if one vehicle has a sudden deceleration, and without sufficient distance you'd end up with one spectacular pileup.
I'd hate to use such a system in the rain. The road is worn down (i.e. is lower) where the tires usually run. That means the pavement is drier if you drive on a "rain line" that is slightly offset from the usual line. This is often enough to keep the car from hydroplaning.
Imagine this: You enter a major arterial roadway -- I-95 or Route 80 and you program your destination -- an exit #, a rest area, etc. Your car joins the pipeline, syncs up with all the other cars on the road, accelerates to a (fast) cruising speed (100+ mph), and you sit back and enjoy the ride. No lane-changing, no "human" failures (like slowing down just because the sun changes direction) - and when it comes time to arrive at your waypoint, the system safely gets you out of the flow and into a "manual" lane. Then you take over from there.
Could be awesome...