It is complex and time-consuming for obtaining accurate dynamics of industrial robots. The flexibility of the joint not only increases the adverse effects of nonlinear factors, but also makes the controller design of robot significantly difficult. In order to improve the performance of robots containing joint flexibility, cooperative learning (COL) theory is proposed in this paper. Based on this theory, a model-free intelligent controller is trained and successfully applied to the flexible joints. Compared with the conventional PID controller and RBF controller, the cumulative tracking error produced by the cooperative learning intelligent controller is reduced by 38.25% and 31.08%, respectively, and the robustness is improved in the trajectory tracking experiment and the robustness experiment. The experimental results demonstrate the feasibility and effectiveness of the cooperative learning theory.

No abstract

No abstract