A Novel Path Error Compensation Method for Robotic Milling by Using Hybrid Temporal Network

In robotic milling, the path accuracy of the industrial robot is a crucial factor in ensuring the contour accuracy. Most existing compensation methods primarily focus on the single-point accuracy of industrial robots while neglecting the continuity of the milling path. This poses challenges to ensure the contour accuracy of the workpiece. Therefore, this article proposes a novel path error compensation method for robotic milling, first introducing a hybrid temporal network (HTN) to predict path error and then using a path iterative compensation (PIC) method to compensate for these errors. The HTN processes spatial information of milling path through convolution operations with error similarity of the industrial robot. Meanwhile, the temporal relationship of milling path can be extracted by temporal layers. Moreover, a dataset augmentation method is proposed to ensure the HTNs performance by comparing the sequence dispersion of each training milling path. Based on the HTNs prediction results, the PIC method is proposed to improve the final path accuracy by iterative compensation. To show the feasibility of the proposed method, a series of experiments are conducted. It is proved that the contour error of skin parts is within ±0.35 mm.