SIMPLE ROBUST SCHEMES FOR CARTESIAN SPACE CONTROL OF ROBOT MANIPULATORS

Direct Cartesian space control of the robot end-effector requires a controller to perform online computations of the inverse robot dynamics, forward kinematics, forward and inverse Jacobians, and time derivative of the Jacobian. The complexity of these computations generally exceeds the capabilities of present-day microcomputers for high sampling rates. Additionally, inverse dynamics requires a highly accurate robot dynamic model, which is difficult to obtain in practice. Model uncertainties can cause unacceptable performance degradation. This paper presents two controller design schemes that substantially reduce the computational load as well as robustify the controller. Such controllers make microcomputer control of industrial robot manipulators practical. Simulation results of the proposed controllers for a PUMA 560 arm are presented.