I was part of a study, somewhat along those lines, in the late 80's. The systematic lodging of movement at various points - radius/elliptical/travel lineage and so forth, had so many complexities to it/them that the project was tabled. The human element with respect to size and physique/physical ability/fatigue rates and variables associated with fatigue, required a deviation of the mean for every motion being factored. (at least that's what I was told) To say it was a project, in a micro sense, was an understatement. The idea in the beginning was to issue a paper on the major human forms of size, age and competitive intensity of the human body in total and in part(s) - the upper torso, shoulders and arm(s) measurements and projections similar to your remarks. Like I mentioned above, the thing got so complicated that the study was tabled.
However, all that was back in the 80's. Today's computers and model makers can, I'm sure, make short work of all that detail and exceptions/tangents. Good luck with your project.
I would like to state though, in my opinion, the variability with the human condition(s) would be a challenge to say the least when compared to the repetitive characteristics and tendencies of the robotic design - would it not? How would you factor in variances like balance inequalities passed up from the lower body to the torso, shoulders and arm(s)? In addition, fatigue and soreness from prior functions is a human experience not usually associated with robotics.
Fascinating topic though. I wish you success with your study.