Abstract: Industrial robots have high repeatable positioning accuracy but their absolute positioning accuracy is relatively low, which limits their application in high-precision machining. Traditional methods to improve absolute positioning accuracy include geometric parameter calibration and off-line error compensation, but their absolute positioning error is hundreds of microns and cannot meet the requirements. Online compensation is a more effective method, but most of the relevant researches are based on simple PID control, which is difficult to achieve accurate trajectory tracking in complex conditions. Thus, a high-precision robust control method is proposed to improve the control precision and robustness in online compensation. The method uses a laser tracker to measure the robot's terminal position in real time and identifies the external disturbance through a second-order disturbance observer. On this basis, a sliding mode controller is designed to complete the robot's robust control. The boundness of observation error of disturbance observer and the asymptotic stability of closed-loop system are proved by Lyapunov method. The proposed method is validated on a COMAU robot, experimental results show that the root-mean-square value of tracking error module in Cartesian space is 0.037 mm when the proposed control method is used for online compensation, which is only 39% of PID based online compensation method.

Key words: industrial robot, online compensation, disturbance observer, sliding mode control