Here’s a drawing that shows the basic structure I have been working on. There are 5 drive wheels which are self-turning “wheel-motors.” The figure on the left shows how many wheels it takes to do the job (although 3-wheel sets may be substituted for 4-wheel sets on turning and guide wheels, with minor loss of stability, just as a three-legged table or a 3-wheeled car is possible but not as stable). Note that half of them become inactive in the process of switching tracks (3rd figure). In the second and third figures the red “right turn” wheels are in the upward, engaged position, allowing all of the wheels on the green “left turn” side to disengage. The ability for cars to do the switching themselves, instead of having to build many switches in the tracks (like a railroad) is pretty much a standard feature of all modern PRT designs. Keeping all of the wheels aiming parallel to the track even as the track turns sharply is the challenge, although such tight cornering would only be for very low speeds anyway. Nonetheless, any good designer would want to reduce such frictional losses and associated wear and I am no exception. If the wheels seem very bunched together it is because I originally drew this as part of a 2 assembly set, much the way train cars have two separately pivotable wheel assemblies per car. These assemblies, connected by a universal joint, would enable extreme flexibility in track layout including those very tight turns I referred to earlier.
Addendum – I wrote and drew that post while still up in the woods of New England, and have since spent some time at RIT (Rochester Institute of Technology) hooked up to broadband, so I have had a chance to further my education (via online video tutorials) on what I consider to be a pretty exciting development, a free 3D modeling program from Google. So here is the extent of my abilities so far. Here I have experimented by using the “3-wheel sets” that I referred to above. In this one the green wheels are in the engaged position and the red ones are down.
