I think that the whole complex of tasks related to the solution of the manipulator's inverse problem can be performed directly in Maple. It will be faster and even more convenient than in specialized CAD. Some known schemes of manipulators are ready for testing; it remains only to perform the formalities associated with the recording of control programs.
1, It is clear how to make a mathematical model of a manipulator. And for this there are several examples.
2. The problem of finding the starting points of motion is conveniently to perform using Maple graphics and with solve or fsolve functions, for example, in a separate program.
3. The solution of the direct problem is simply routine programming.
Of course, I may be wrong, but for some reason it so seems to me.