Increasing Trajectory Tracking Accuracy of Industrial Robots Using SINDYc

Abstract

Abstract In this work a feedforward control approach based on SINDYc (Sparse Identification of Nonlinear Dynamics with Control) is proposed for increasing the trajectory tracking accuracy of industrial robots. Initially, the dynamic relationship between the desired and the actual trajectory is sparsely identified using polynomial basis functions. Then a new trajectory is created from the desired trajectory using a feedforward controller based on the inverse of the sparsely identified dynamic model. The effectiveness of the proposed approach is evaluated by a simulation study in which 4 different KUKA robots were tasked to follow 16 distinct trajectories based on ISO 9283 standard. The obtained results show that the proposed method successfully models the dynamic relationship between the desired and the actual trajectory with accuracies above 98.09% when all of the robots are considered. Moreover, the developed feedforward controller improves the trajectory tracking accuracy of industrial robots by at least 91.1% and 94.5% for position and orientation tracking, respectively while providing parsimonious models.