Configuration optimization of robot in-place machining system based on error prediction surrogate model

Industrial robots have become another important manufacturing equipment in addition to machine tools due to their advantages such as large working range and flexible working modes. However, compared with machine tools, larger errors restrict their application in precision-dependent scenarios. The robot's processing error is directly related to the system configuration. Under the same task, different system configurations will directly lead to different processing qualities. In the task scenario of in-place processing, the external configuration of the robot is under controllable change. Under such circumstances, how to reasonably and reliably select the robot processing system configuration and reduce the robot's processing error needs to be paid attention. Existing research carries out optimization based on theoretical error prediction models, and lacks the problem of optimization accuracy caused by differences between theoretical models and actual machining error prediction models. To this end, a configuration optimization strategy for robot in-place machining system based on error prediction surrogate model is proposed. In the proposed method, based on the spatial sampling strategy, a data-driven robot variable space processing error surrogate model is established, which realizes accurate prediction of the robot's processing errors in different states. On this basis, a robot processing configuration optimization model is established to realize the optimization of robot processing configuration. Through experimental verification, the constructed surrogate model achieves the optimal prediction performance of 0.009mm average absolute deviation, and the optimized average robot processing error is reduced to 0.194mm, which provides a good system foundation for subsequent compensation and control of robot processing errors.