>>14266213that's something I would be interested also but I don't have the time now, nor the maths skills yet
but it's very complex
first you have to get the degree of liberty of each articulation isn the body, and there isn't a single articulation which is nice and square, they all have weird shapes so it's more complicated than textbook exmaples of degrees of freedom of mechanisms
ie, you need differential geometry
so that mean you also need 3D images of each articulation
and when you have this, you also need to add tendons, ligaments, and muscles
which also have a geometric aspect plus some complicated properties. They are elastic. And the degree of elasticity probably vary upon how much you are stetching / binding the articulation bound to the given ligament or tendon or muscle
Also, for a given person, the mechanical property of its own join, ligaments, tendons and muscles will vary from day to day, depending on how much the person has sit on its ass and what exercises he has done, what muscles he has stretch, reinforced, in what intensity, etc..
Si I think a perfect model would start with the geometry (which gives degrees of liberty) of joints and bones. This would be static.
And on top of that, you would add muscles, tendons and ligaments with their mechanical properties (which would be ideally varying accross that said ligaments, think about a heat map). And all this would be dynamic.
The dream, then, having all this would be to obtain your own (if you want to model your own body or someone else's) mechanical properties on a given day. I think you could do this by combining several data sources. Like doing a fixed set of movements and being recorded on video, and from the video, you could infer a lot. Plus putting proprioceptive sensors all over you body.
Now if you want to simulate a human moving (not recording its movement patterns), you have to take into account neural aspect, activation patterns, etc.. With some complicated geometric+temporal pattern
