Load Estimation of Manipulator Based on the Optimal Sequence of Exciting Poses

The gravity compensation method is widely used in the robot system composed of the links and the rotating structures. The replacement of the robot end-effector brings uncertainty into the compensation model. For this problem, an offline identification method of load parameters based on joint torque and position information of the robot is proposed. Based on the robot statics method, two calculation models of load parameters are proposed, and the least squares solution of the load parameters is obtained by collecting the joint torque and position information of the manipulator under multiple static poses. Furthermore, the problem of choosing the poses for identification is studied, and a multi-objective optimization problem is proposed to guarantee the accuracy and simplicity of identification simultaneously. The multi-objective particle swarm optimization method is used to obtain the optimal poses for identification. According to the identified load parameters, the calculation method of the load gravity compensation for each joint of the manipulator is given. The experimental results show that the proposed method is of high identification accuracy. The minimum value of the identification error of the load mass is 0.007 06 kg, and the maximum value is 0.151 kg. The minimum value of the identification error of the load centroid position is 0.025 4 m, and the maximum value is 0.122 m. The feasibility and effectiveness of the above method are verified.