hmmm it sounds really interesting, it could be a hard but exciting "game".

strangely, sometimes I do something similar, but unfortunately I can't separate my brain into 2 parts, when developing new pathfinder+movement solutions (ahh they are never perfect, even perfect has no meaning in this context), I usually just read the theory and watch presentations, but normally do not check codes if available, what makes the whole thing more time consuming, but the final result is really 100% own work. maybe the 2 only exceptions are A* algorithm, and the very fast hybrid array + linked-list approach in A* data handling, made by Marco Pinter...