An Adaptive Inverse Kinematics Algorithm for Robot Manipulators

This paper presents an adaptive algorithm for solving the inverse kinematics problem for robot manipulators. The algorithm is derived using Model Reference Adaptive Control (MRAC) theory, and is computationally efficient for on-line applications. The proposed scheme requires no a priori knowledge of the kinematics of the robot if Cartesian end-effector sensing is available, and requires knowledge of only the forward kinematics if joint position sensing is used. Computer simulation results are given for the redundant seven DOF Robotics Research arm, and demonstrate that the proposed algorithm yields accurate joint angle trajectories for a given end-effector position/orientation trajectory.