Online compensation strategy based on set-membership filtering for manipulator machining

In manipulator machining, the motion accuracy of each joint is a core element to guarantee the machining precision of the end-effector tool. To address the degradation in joint tracking performance caused by the limited capability of feedforward control in fully compensating for tracking errors, this article proposes an online compensation strategy based on set-member filtering (OCS-SMF) for manipulator joints. The proposed strategy employs set-member filtering to accurately estimate the tracking error at the upcoming sampling time. Following this estimation, a multi-stage filtering structure is embedded to generate compensation signals online through real-time iterative computation. These compensation signals are directly injected into the feedforward signals to enhance the motion accuracy of the joints. Finally, the experimental results demonstrate that the proposed strategy improves joint tracking performance in several manipulator machining tasks. Specifically, for the point-to-point tracking task without disturbances, the proposed method decreases the MA (the move average) and MSD (the moving standard deviation) metrics by up to 21.60% and 33.3%, respectively, compared with the subsequent comparison methods. For the point-to-point tracking task under disturbances, the proposed method achieves maximum reductions of 21.05% and 31.25% in MA and MSD, respectively, compared with the subsequent comparison methods. For the quasi-sinusoidal tracking task, the proposed method decreases the MA and MSD metrics by up to 25.10% and 20.1%, respectively, compared with the subsequent comparison methods. For the irregular sinusoidal-like tracking task under disturbances, the proposed method decreases the MA and MSD metrics by up to 30.60% and 20.29%, respectively, compared with the subsequent comparison methods.