A Novel L₁-and-L₂-Norm-Integrated Parameter Identification Model for Robot Calibration

Robots promote the social development by enhancing production efficiency, reducing labor costs, and improving service quality. Meanwhile, their application fields such as healthcare, transportation, and education can drive technological advancement and innovation, thus fostering economic growth and improving our quality of life. However, due to mechanical wear from prolonged operation, the robot absolute positioning error can reach several millimeters, thus it is impossible for them to perform precise tasks. To address this intractable problem, this study innovatively proposes a robot calibration method (FPSA) combining the search algorithm based on proportion integration differentiation (PID) controller and fuzzy strategy for adjusting hyperparameter, its novelties include: 1) designing a novel L₁-and-L₂-norm-integrated parameter identification model, then search algorithm [PID search algorithm (PSA)] based on PID controller is adopted to solve this model, which achieves the accurate identification of robot kinematic errors; and 2) developing an efficient fuzzy strategy for adjusting hyperparameter to achieve hyperparameter adaptation in the robot parameter identification model, thereby effectively enhancing the calibration computation efficiency and accuracy. Moreover, we conduct extensive calibration experiments on an HSR JR680 robot. These experimental results demonstrate that compared with other advanced calibration algorithms, the maximum error of the proposed FPSA calibration method is 9.57% higher than that of the most accurate identification model solved by PSA algorithm, which has single L₂ norm regularization and objective function. Therefore, this research provides excellent calibration service for a robot, thus contributing to the prosperous development of application fields such as industrial manufacturing and smart agriculture.