I've been thinking about torque motors, which are beautiful and interesting gadgets, and specifically about simulating them to a FADEC. Being lazy, I do my hard thinking while I'm asleep. Coffee and a shower provoke delivery.
I can't build a reasonable board that would absorb energy during acceleration and return it during deceleration, but most customers wouldn't demand that. But we could simulate the system moment of inertia and loading and compute angular position and report that and then simulate an encoder.
We can have programmable coil resistances too. Programmable inductance is a whole nother issue.
COE is a PITA.