Flutterby™! : Circuits, model railroads and snakes

Flush with my microcontroller successes and barely coherent due to my lack of sleep from the trip to deploy it, I've been searching for a few things. Looking for information on current sensing lead me to Rob Paisley's model railroad electronics pages, specifically his various transistor throttle circuits go step-by-step through the sequence so I can understand the whys and wherefores of each component. But in my searching I've also learned that knowing a lot of this is like knowing assembly language, especially since I'm planning on driving everything from a microcontroller and software anyway: it seems perfectly reasonable to just get something like a LMD18200 3 amp H-bridge motor driver with internal overload protection and a current draw output and not worry about designing my own circuit breakers.

Relatedly, I'm looking for a good practical intro to electromagnetic actuators. I've been thinking about ways to build a cheap robot that goes beyond just repurposing an RC car chassis or similar. The big problem is that as you add axiis of motion, prices go up. Yes, I could build yet another hexapod that shuffles like a tank, but why? And doing a hexapod right takes 16 motors or more. I could do a two legged walker that shuffles in 3 or 4 axiis, but the 11 or 12 (3 ankle, knee, two hip, times two) to do it right starts to cost a lot.

What about a snake? Drop a small wheel on each segment, drive the joints between each segment with two electromagnets and a fixed magnet. Driver circuits become easy, especially if the stall currents on the actuators is low. If it's long enough I'd bet it'd be pretty good at some terrain that'd give anything else of equivalent articulation a lot of trouble. Scales nicely, could be built small or large. Probably not something I'll get to soon, but worth just letting a few background neurons play with the idea a bit more.