Kinematic Calibration of a 5-DoF Parallel Machining Robot with a Novel Adaptive and Weighted Identification Method Based on Generalized Cross Validation

• A novel adaptive and weighted identification method based on generalized cross validation (GCV) is proposed for kinematic calibration of a 5-DoF parallel machining robot, addressing ill-posed identification issues. • The method incorporates Gauss-Markov estimation and prior physical information to construct a weighted identification model and cross-validation function, improving numerical stability and accuracy. • Experimental results show significant error reduction: maximum position error from 2.279 mm to 0.028 mm and orientation error from 0.206° to 0.017°, outperforming typical least squares methods. • The kinematic error model considers non-ideal constraints and screw self-rotation, capturing the complex error characteristics of the parallel robot.