LKF Based Robust Position Control of Robotic Manipulators

A new control strategy for robotic manipulators is proposed in this paper. The position control system design is based on a joint local PD control technique with a feedforward compensation term. The PD controller is designed considering a flrst-order linear reference model for each joint. The controller gains are tuned on average values of the inertial moments, evaluated in the range of the allowable configurations of the manipulator links, and on nominal viscous friction coefficients. To obtain the specified ideal dynamic performance of the robot, the feedforward signal has to compensate for the equivalent disturbance rising from the dynamic coupling among axes, centrifugal, Coriolis, and gravity forces, unknown and timevarying external load torque, non-linear friction torque terms, and parameter variations occurring in the manipulator dynamics. The robustness of the proposed control system is guaranteed by the recursive algorithm, based on the discrete linear Kalman filter (LKF) theory, which has been developed to online estimate the non-linear time-varying equivalent disturbance. The LKF also gives the estimates of position and speed that are used as feedback signals for the joint control system. The control algorithm has been tested on the main three axes of a COMAU SMART 3S industrial manipulator.