Robust Trajectory Tracking Method for Flexible Joints Based on Sliding Mode Observer

Traditional control methods for flexible joints are susceptible to errors caused by irregular parameter variations and external random disturbances. This paper proposes a robust control method for flexible joints without torque sensors and employs an improved least squares method for dynamic parameter identification. Firstly, based on the traditional mathematical model of flexible joints, a torque controller is designed, expressing the reference torque as the sum of the trajectory error compensation and the observed torque. Secondly, irregular parameter variations and external disturbances are treated as total disturbances and a disturbance sliding mode observer (DSMO) is designed for torque estimation, with the convergence of DSMO proven. Finally, considering the influence of non-invertible and ill-conditioned matrices on parameter estimation in traditional least squares methods, the ridge regression method is employed to identify the dynamic parameters of flexible joints. Experimental results demonstrate that the proposed method exhibits a good dynamic response, effectively reducing the impact of irregular disturbances during human–robot interactions. Both the speed and robustness of the system are significantly improved.