A modular approach to redundant robot control

This paper describes a modular approach for computing redundant robot kinematics. First some conventional redundant control methods are presented and shown to be "passive control laws", i.e. they can be represented by a network consisting of passive elements. These networks are then put into modular form by applying scattering operator techniques. Additional subnetwork modules can then be added to further shape the motion. Modules for obstacle detection, joint limit avoidance, proximity sensing, and for imposing nonlinear velocity constraints are presented. The resulting redundant robot control system is modular, flexible and robust.