PCRR, I think GRJ cut to the meat of it, but as I tend to do, I feel like rambling on with technical stuff. The idea is great. As the engine gets near a switch you press a button and throw a switch. With the fewest number of parts here's what you need to do that.
A TMCC base to talk to the switches on the 455KHz radio frequency that they use.
A sensor of some kind on each leg of each switch to detect that a train is approaching. I would probably use an RFID tag, though the infrared sensors that Lionel already makes would work as well. In any case the engines need to be fitted with some form of electronic id tag so that the switch knows not only that an engine is coming, but WHICH engine is coming.
Sensors/connections to the switches to know which way they are thrown.
A 2.4GHz radio that can listen for commands from the remote.
All of these sensors, the radio, and the TMCC base then need to be connected to a micro-processor of some sort that can take the radio command when you press the remote button, scan the sensors to see if that engine is near a switch, then if it is, look up the TMCC address for that switch, find out which way the switch is thrown, then send out the command to throw it the other way.
This sounds like an awesome science fair project, but it is impossible for a marketable product. No one is going to want to to add a ton of wires and sensors to switches where the whole point is that they require no wires at all. Never mind that the cost per switch would be insane. I suppose you could use the 2.4GHz wireless network for all the sensors as well to get rid of the wires, but you still need to add 4 sensors on each switch
As an afterthought, you also need some way to program the switches into the device, so either a way to read the Cab1, and use that for programing, or a keypad on each switch controller.
Afterthought #2: Even if you have all this in place, you're still humped if you want to install back to back switches, as there would be no way for a device to know which switch to throw. I suppose you could go with high precision GPS, but I believe that the sort of 1-2 inch precision that would be needed would make legacy engines seem like they are in the bargain price bin.
Afterthought #3: After all that, you are still not compatible with and current engine as you would have to install the tracking hardware on every engine you have.
Now If I wanted to make an entry level LionChief switch controller, I would sell a unit with a similar remote to those used in LC/+ engines. Instead of the throttle, each 'click' of the knob would point to a number (1-32, say), and that would select TMCC switch address 1-32. A box would connect much like a TMCC base and talk to that remote, letting it throw TMCC addressed switches. I doubt this sort of product will come to market because it is firstly, too complex for the target audience, and second, more complex than just using a cab1. I suppose it is possible that Big L could release LC-Command switches that could talk directly to a 'switch remote' as described above, but I really, really doubt it.
JGL
Currently waiting on senate approval for my nomination as ambassador of LionChief.
In order to handle the command switches, you have to individually address them with a TMCC command coming from a TMCC/Legacy base. This opens up the whole can of worms we've been talking about! 
. It's bizarre but predictable that any discussion about increasing the capability of the relatively new LC/LC+ systems, or future enhancements brings out the Luddites.
