关键词: 工业机器人, 末端负载, 扭矩平衡, 参数辨识, 力矩投影

Abstract:  To address the challenges of decoupling center of mass parameters in existing end-payload identification methods and the difficulty of implementation on robots with non-open controllers, a torque-balance-based modeling and identification method was proposed for robot end-payloads. The identifiability conditions of the end-payloads were analyzed under joint torque balance, and identification models for the end-payload mass and center of mass position were established. To further decouple the mass and center of mass parameters, a three-step identification strategy was designed, where the load mass was identified first, followed by the center of mass position in x and y, and finally in z. This strategy effectively eliminated the error terms introduced by the projection of joint torques in the identification models. The efficiency of the proposed method was validated through simulation and experiments. Compared with the built-in identification method of a non-open-source six-degree-of-freedom robot, the average error in mass identification is reduced from 0.103 kg to 0.032 kg, while the average error in center of mass position identification is decreased from 50.25 mm to 4.14 mm. Furthermore, compared with dynamics parameter identification, the mass identification error is reduced from 0.179 kg to 0.083 kg, and the center of mass position error is reduced from 10.13 mm to 4.33 mm.

Key words:  , industrial robot, end-payload, torque balance, parameter identification, torque projection