An unsupervised prediction of robotic machining error for new tasks under historical tasks knowledge distillation

Contribution

In this paper, we propose an unsupervised method for predicting robotic machining errors in new working conditions, based on historical task knowledge distillation. First, a robust historical model is trained using abundant historical data. Then, a lightweight student model is developed to distill and transfer knowledge from the historical model, focusing on three key aspects: accuracy, model structure, and data. The proposed method was tested on two new tasks with significant parameter and shape differences. The results demonstrated the substantial advantages of the knowledge distillation approach for unsupervised prediction in novel tasks.

Recommended citation: T. Zhang, F. Peng, X. Tang, Z. Yang, R. Yan, An unsupervised prediction of robotic machining error for new tasks under historical tasks knowledge distillation, Procedia CIRP, 133 (2025) 161-166, https://doi.org/10.1016/j.procir.2025.02.029. https://doi.org/10.1016/j.procir.2025.02.029