Calibration and error compensation of the intelligent robot welding system for the milling rotor

The spatial pose accuracy of tool holders on a milling rotor directly affects the operational quality and service life of cold milling machines. Challenges in accurately positioning and installing these tool holders arise from the large size of the drum workpiece, limited direct measurability, and multi-parameter coupling in multi-robot systems. This study establishes an intelligent robotic welding system based on a milling rotor experimental platform, integrating key modules such as ground rail parallelism analysis, global coordinate system construction, laser-based feature fitting, vision-guided grasping, and dual-mode Hand-Eye calibration. A hybrid calibration approach combining Eye-in-Hand and Eye-to-Hand configurations enables high-precision acquisition and transformation of six-dimensional spatial pose information for both the rotor and the tool holders. To address the nonlinear coupling of pose errors, an Improved Particle Swarm Optimization–Support Vector Regression (IPSO-SVR) model is further proposed for grouped modeling and compensation of positional and angular deviations. Experimental results demonstrate that the proposed system architecture and calibration methods are reliable and effective, with the compensation model significantly enhancing pose accuracy and welding consistency.