In the meantime, we'd do better to exploit other improvements. Once LEDs or an equivalent become cheap enough, not only does it mean we can vastly increase the amount of roadway illumination (arrays and strips rather than relatively few point sources, as we have now with ordinary lamps and luminaires), which means a greater energy use and cost, but a vast improvement of our roads and their use in exchange; we can also do something as mundane but much more valuable than any fancy automation gizmos, namely embed LEDs or other future small lamps in the roadways to aid visibility during darkness and inclement weather. (We could even have color changes and aimed lamps in the case of reversible-flow lanes, don't forget, which is a good way to improve usage of scarce capacity in metro areas that have commuter flow patterns.) This is already an idea considered to be used for crosswalks:

http://www.tfhrc.gov/pubrds/04jan/03.htm

It's even an alternative to overhead lighting, in particular, in rural areas as well as where there are environmental problems with overhead or more-intense illumination, as in this example:

http://www.dot.state.fl.us/research-Center/Comp…

http://www.airporttech.tc.faa.gov/Safety/Downlo…

But only if they make it legal to pee out the window when the rest of the platoon says “you shoulda gone befor we left!”

1993:

http://www.ahmct.ucdavis.edu/images/UCD_ARR_93_…

San Diego and other experiments:

http://www.path.berkeley.edu/PATH/Publications/…

http://www.path.berkeley.edu/PATH/Research/demos/

http://www.path.berkeley.edu/nahsc/pdf/AHS-Tech…

And, to address your thought,

“I’ve long thought we could turn those reflectors that demarcate lanes on roads into buoys read by [...]“

Well, here you are.

“This report investigates the feasibility of using the magnetic markers, already present for lateral vehicle guidance, for longitudinal control. Instead of using a “vehicle-follower” method implied by relying on direct inter-vehicle distance measurement, the use of the magnetic markers gives a “point-follower” scheme, relying on a fixed reference system. …”

http://www.path.berkeley.edu/PATH/Publications/…

P.S. Here's the lateral-control portion.

http://www.path.berkeley.edu/PATH/Publications/…

This isn't really necessary, and shouldn't be required, along with other techno-gadgetry like automatic brakes to prevent tailgating or rear-end collisions — nice to have, but overhyped and not essential — and adding unnecessary expense. Certainly it's no substitute for driving skills.

“Do you guess that if we had to do it today, we’d have the wherewithal to build the Interstate Highway System?”

Today's Congress and Executive Branch would be little better suitable for this than for the lunar landing.

And worse still, such an undertaking would be hampered by anti-auto and environmental activists.

2. A better idea, which is suitable for electric vehicles someday for inter-city travel, would be to put the vehicles under positive system control (automation), which would permit respectably high speeds (such as 250 miles per hour), and which might involve platooning to lower air resistance and energy costs of such a system, but which would favor individual, independent vehicle operation (putting them into platoons or “trains” as and when merited), an inter-city real-world-speed kind of system resembling some long-ago concepts of Personal Rapid Transportation (with individual vehicles) in urban areas.