Digital human modeling can nowadays be animated by means of motion capture systems and inverse kinematic algorithms. This process of animating digital models is called motion reconstruction. It consists of calculating the joint angles corresponding to the kinematic architecture of a model. This process can be seen as an optimization process, minimizing the distance between measured and reconstructed marker positions.
For lower body, upper body and head segments, this process can be easily over constrained seeing as it is experimentally possible to put at least two or three markers on these bodies. On the other hand, the spine is often modeled as several body segments and markers cannot be placed on each of them for the simple reasons that subjects sit in a car and that it might be decomposed into very small body elements. Under these conditions, an infinite set of spine posture can achieve the same constraints on the pelvis and torso. For example, if the spine is compressed, the spine posture solution can be flexed, arched or folded into pleats.
The objective of this paper is to present a method that will guide the reconstruction process towards physiological postures at each frame of the motion. For one set of constraint conditions, only one posture should be selected by the reconstruction process and spine degrees of freedom should no longer be independent from each other. This is the reason why this method is said to be coordinating the degrees of freedom. Then, the problem is to define the coordination law that will drive the reconstruction procedure. Experimentations were therefore conducted in order to evaluate and validate a coordination law for spine degrees of freedom.