Two stage control for high degree of freedom articulated figures

Control of high degree of freedom articulated figures such as robots and simulated human like models in an artificial environment is a difficult problem. The endeavor has been referred to as “teleological modeling” [7] (pp 315–321). With teleological modeling an object is described by a collection of desired goal (Greek “teleos”) states over time. The modeling paradigm achieves these states under certain constraints such as obeying the laws of physics. The current work examines a viable solution to the teleological modeling problem. The problem is broken into two stages. First the global motion of the articulated figure, as determined by the center of mass trajectory, is found by optimizing an objective function including physical constraints. The joint space trajectories required to achieve the motion are found using a separate optimization. Specifically the technique of resolved motion rate control is used. In this technique the pseudo inverse of the Jacobian performs a quadratic optimization on the joint velocities. Trials show that realistic motion is obtained for a 17 degree of freedom articulated.