Draft Post on Self-Driving Cars
I am writing this quickly, so expect it to undergo some edits and updates later.
Google's self-driving cars have been in the news recently (I'll add a link later). While the AI involved in making a car capable of driving itself among human drivers is impressive, it's only the very first part of the puzzle and, ultimately, should only be an emergency fallback. Here's a vision of a future of self-driving cars based on operations research, successful business models, and government-industry partnership.
Consider a platform car. This has been discussed elsewhere (need link), but the basic idea is that the wheels, gearing (if any), motors, etc. are all built into a platform with a (nearly) flat surface on which a passenger compartment can be placed. The compartment sits securely on the platform and control signals for the resulting complete vehicle pass through an interface between the compartment and the platform. The power source can either be part of the platform or part of the compartment, depending on desired economics. For now, let's assume the power source is part of the platform rather than the passenger compartment, and the owner of the platform (who probably isn't the owner of the compartment) is responsible for keeping it fueled. The opposite results in a different economic model which is probably less vulnerable to monopolistic abuse, but it's more difficult to discuss.
Now imagine a world in which everyone who would otherwise own a car actually owns a passenger compartment. They are responsible for keeping it fueled, but the platform on which it is sitting (say, in they garage) is rented rather than owned. More to the point, it's interchangeable.
A typical commute might involve driving to a local commuting hub (perhaps equivalent to a highway entrance) at which the passenger compartment is removed from its current platform and attached to a platform that is driven by either a centrally coordinated server or by a peer-to-peer federated server running on all the platforms. The new vehicle automatically merges into high-speed traffic, other cars automatically making room for it. The traffic can move at speeds that human drivers couldn't be trusted to navigate safely, in large part because the system is highly predictable.
The passengers will have programmed their destination into their compartments, thus transitioning to another highway will be automated as well, and the passengers won't have to worry about missing their exit. If the next roadway is owned by the same entity (company or gov't), or the owners have appropriate agreements and technical compatibility in place, the vehicle won't even require a platform change.
When the vehicle nears the passengers' destination(s), they will be alerted (they might have dozed off, after all). They will have to verify that they are prepared to take control of the vehicle and, when it exits the highway for surface roads, it will be placed on another interchangeable platform. The passengers (or, rather, the driver among the passengers) will drive to their destination.
This vision has a variety of advantages, not least of which is that it can be transitioned gradually. The only government mandate required is that all new cars conform to the compartment/platform standards. Automated roadways can spring up wherever it is economically feasible, and can be funded/managed by private or public concerns as appropriate. I've glossed over fueling and payments in this discussion, not to mention how it interacts with trucking, rail, and even air travel, but that's why this is a draft. I'll get to it. In the meantime, take this as a sketchy vision of a viable future of improved commuting and ground transportation.