While our inverse kinematics system shows promise, additional work could make it much more realistic and usable.
Our heuristic estimation of which positions are easier or harder is very limited. A better approach would be to define a realistic comfort function based on muscle properties, as well as a separate energy cost function that measured the work to go from one position to another in a set amount of time. A complete model would include the effects of momentum, angular momentum, and gravity.
Furthermore, our methods for solving these constraints are slow and sometimes get stuck. Faster and more robust optimization methods could make a big difference in the usability of this system.