Kinematic self-identification of a redundantly actuated parallel manipulator using differential evolution

The self-identification of a redundantly actuated parallel manipulator is transformed into an optimization problem, and then differential evolution algorithm is used to obtain a globally optimal solution of the kinematic parameters. Based on the kinematic equations of the parallel manipulator, a new optimization function is formulated by eliminating the passive joint angles and decoupling the kinematic parameters. In order to obtain the global optimum, differential evolution algorithm is applied to minimize the optimization function. The proposed self-identification method is realized on an actual parallel manipulator, and the comparison on the actuated joint errors indicates that the identified kinematic parameters are much more accurate than the nominal parameters. Finally, the kinematic control experiments are carried out, and then the tracking accuracy of the end-effector based on the identified kinematic model is compared with the results by using the nominal kinematic model.