A Method to Improve the Accuracy of Hand-Eye Calibration Based on Data Filtering and Parameters Optimization

Hand-eye calibration is a critical step for camera-robot systems. Addressing the problems of low accuracy in hand-eye calibration, the paper introduces a method to obtain the transformation matrix based on data filtering and parameters optimization. In constructing the calibration equation system, certain equations exhibit larger errors due to factors such as the angles, positions, and noise when the camera captures the calibration board. Additionally, following the acquisition of the initial calibration solution, the traditional methods of optimizing calibration parameters frequently based on two-dimensional. However, the calibration results need to ensure the accuracy of the visual-guided robotic arm in 3D space, which leads to a discrepancy between the optimization objective and the actual requirements. Based on this, the proposed method can be divided into two steps. Firstly, the calibration equations are filtered using Random Sampling Consistency (RANSAC) to reduce the calibration data collection error. Subsequently, the spatial distance error serves as a constraint to iteratively optimize the calibration parameters, and the error is used as an evaluation metric for calibration quality, thereby enhancing calibration accuracy further. Finally, simulation and real experiments were conducted to validate the precision and robustness, and the calibration results were applied in robotic arm pick-and-release operation to evaluate the practical application effectiveness. Experimental results show that the proposed method has higher accuracy and stronger robustness compared with other traditional methods.