Neural Network based dynamic trajectory tracking of Delta parallel robot

This paper proposes a geometric method to solve the forward kinematic of the Delta parallel robot and then study the workspace. The inverse kinematics is also presented as the prerequisite for the dynamics. A simplified dynamics model is built by using virtual work principle. As a result, computed torque method based controller is obtained, which needs the accurate parameters of the model. But high coupling and nonlinear properties of the dynamics model is so explicit that it becomes a major impediment to development. So Neural-Network based controller is proposed, in this paper, to compensate errors caused by the uncertainties of the model´s parameters. A joint simulation is carried out for the computed-torque-based controller and Neural-Network-based controller. The results show that the performance of the proposed controller is prominent when the end effector tracks a given trajectory.