Research on Non-Probabilistic Reliability Partitioning Methods for Large Workspace Robots

摘要

Large-scale high-end equipment robotic precision operations have large workspaces and numerous uncertainties. They have an unevenly distributed error effect on the position in space. The current conventional stereotyping method ignores the uncertainty in the robotic system; the probabilistic or fuzzy method is often due to the lack of statistical samples in the project, and it is difficult to accurately define the probabilistic or fuzzy model because the probabilistic distribution pattern or fuzzy affiliation cannot be known in advance. In this paper, we propose a non-probabilistic reliability-based robot workspace partitioning method that only needs to know the upper and lower bounds of the values of the uncertain parameters and is adapted to realize accurate calibration of robots in small-sample or information-poor scenarios. The method considers the differences in non-probabilistic reliability of robot end positions in different workspace ranges and uses KLFCM clustering combined with a genetic algorithm to perform a two-stage hierarchical partitioning optimization. The experimental results show that compared with the global compensation, the average values of the upper and lower limits of the x, y, and z direction error intervals of the partitioned compensation are reduced by 31.17%, 7.26%, 14.30%, 34.91%, 2.48%, and 35.82%, respectively, verifying that the method in this study can more accurately realize the partitioned categorization calibration and compensation of the robot, and effectively improve the reliability and spatial adaptability of parameter calibration and compensation of the robot in the full workspace domain. The reliability and spatial adaptability of the parameter calibration and compensation are effectively improved in the entire workspace domain of the robot.