Adaptive Line Trajectory Identification of Industrial 5-DOF Robot Arm Using Neural MIMO NARX Model

Abstract

Abstract. This paper investigates a novel forward adaptive neural MIMO NARX model which is applied for modeling and identifying the forward kinematics of the industrial 5-DOF robot arm system. The nonlinear features of the forward kinematics of the industrial 5-DOF robot arm drive are thoroughly modeled based on the adaptive identification process using experimental inputoutput training data. This paper proposes the novel use of a back propagation (BP) algorithm to generate the forward neural MIMO NARX (FNMN) model for the forward kinematics of the industrial 5-DOF robot arm. The results show that the proposed adaptive neural NARX model trained by Back Propagation learning algorithm yields outstanding performance and perfect accuracy.

Skill learning approach based on impedance control for spine surgical training simulators with haptic playback

“…Furthermore, Φ ( z ) is found by collecting z data while W ^ is found via imposing the back propagation algorithm 38 on equation (10). The objective is to make the haptic “actively” push toward the surgeon’s expected position, thus leading the smallest possible value for the interaction force f .…”

Impedance learning adaptive super‐twisting control of a robotic exoskeleton for physical human‐robot interaction

“…It is known that ϖ can be made artificially small, if l is adequately high. Moreover, Φ i (z) is achieved by collecting z data, which are actual position, velocity and forces of the subject and Ŵ is collected by applying a back propagation algorithm [37] in Equation (11). The goal is to have the exoskeleton robot "actively" push towards the user's target position, hence minimising the interaction force f. Therefore, Ŵ is updated in the direction of the steepest descent relative to the cost function, so…”

“…Moreover, Φ i ( z ) is achieved by collecting z data, which are actual position, velocity and forces of the subject and $$\widehat{W}$$ is collected by applying a back propagation algorithm [37] in Equation (). The goal is to have the exoskeleton robot “actively” push towards the user's target position, hence minimising the interaction force f .…”