Robust control by adaptive Non-singular Terminal Sliding Mode

Abstract

Based on the principles of the Non-singular Terminal Sliding Mode Control (NTSMC), a new control law along with an Unscented Kalman Filter (UKF) has been proposed for robotic manipulators, that can tolerate external disturbances and noises with unknown statistics. First, a neural network module has been used as a discontinuous control part of the NTSMC to enhance the performance of the controller due to chattering phenomenon. Furthermore, a new methodology is proposed which is based on a modified evolutionary algorithm (charged systems search) to estimate the system states by the UKF and the measurement and process noise covariances. To compare this evolutionary method with classical methods, an optimal Unscented Kalman Filter (UKF) algorithm has been introduced that estimates the noise statistics recursively within the algorithm. The proposed control method and observer have been simulated on a 6-DOF robot manipulator.