Caidong Wang,∗ Yongfeng Tian,∗ Huadong Zheng,∗ Hengyuan Hu,∗ and Xin Zhang∗
Multi-joint robot, motion tracking, robot calibration, error compensation
In order to improve the assembly accuracy of the robot, this paper establishes an error model of the robot end effector and compensates for the position in the workspace using the whale optimisation algorithm-extreme learning machine (WOA-ELM) algorithm. Firstly, the workspace is divided into sampling points through grid sampling, and the errors at the sampling points are measured using a laser scanner. The semi-variogram function is used to prove the correlation between the error and the spatial position, determining the possibility of establishing the model. The WOA-ELM error correction model is established by combining the extreme learning machine (ELM) algorithm with the whale optimisation algorithm (WOA). The feasibility of the error correction model is verified on a six-degree-of-freedom robot. The experimental results show that the absolute positioning error of the robot follows a Gaussian distribution in different directions of its coordinate system in space. The absolute error range before compensation is 0.1463∼0.9159 mm, with an average error of 0.4675 mm. After compensation, the optimal absolute error range is 0.1370∼0.3725 mm, with an average error of 0.2182 mm. The WOA-ELM error correction model effectively reduces the ELM algorithm’s randomness, lowers the robot’s absolute positioning error, enhances accuracy and motion stability, and offers strong generalisation performance. This study provides an innovative and practical approach to improve industrial robot positioning accuracy, showing significant potential for high- precision assembly applications.
Important Links:
Go Back
