Nonholonomic and stratified robotic manipulation supplemented with fuzzy control: Theory and experiment

This paper presents initial results in the novel integration of nonholonomic and stratified motion planning, fuzzy control and tactile sensing to construct a robotic manipulation system that is designed to be both dexterous and robust. It is dexterous in that it is fully nonlinear, can explicitly incorporate discontinuities in the motion planning i.e., finger gaiting, and allows rolling finger contacts. The sensing and fuzzy controller are intended to provide robustness that is necessary for real-world manipulation tasks that are characterized by modeling errors and are subjected to unmodeled external disturbances. The method is demonstrated experimentally using a set of four robots with end-effectors equipped with force sensors to provide feedback to the fuzzy supervisory control system.