Hierarchical Feedback and Learning for Multi-joint Arm Movement Control.

This paper presents a general method for hierarchical feedback control of redundant systems, and applies it to the problem of arm movement control. A high-level feedback controller, designed using optimal control techniques, operates on a simplified virtual plant. A low-level controller is responsible for performing a feedback transformation of the physical plant into the desired virtual plant. The method is applied in the context of reaching with two realistic models of the human arm: a 2-DOF, 6-muscle model, and a 7-DOF, 14-muscle model. Simulation results demonstrate the effectiveness of the proposed scheme.