Dynamic parameter identification of robot manipulators based on the optimal excitation trajectory

Abstract

A new method to identify the dynamic parameters of robot manipulator is proposed, which is the identification of dynamic parameters based on Particle Swarm Optimization (PSO). Firstly, the dynamic model including friction of manipulator is established. After the parameter transformation, it is expressed as the linear form of parameter to be identified. Then, the periodic Fourier series is selected as excitation trajectory of manipulator. The combination of condition number, the minimum singular value and maximum singular value of coefficient matrix of dynamic equation is selected as the index of trajectory parameter identification. Particle Swarm Optimization is used to minimize the objective function and obtain optimal excitation trajectory parameters. Let the manipulator to operate according to the predetermined trajectory to obtain the state of motion and torque of each joints. Finally, according the state of motion and torque of each joints, the dynamic parameters of manipulator are identified by using PSO to calibrate dynamic model of manipulator. The result of simulation and experiment indicate that the correctness and feasibility of the identification algorithm is manifested.