Researchers at Hebei University of Technology have developed a payload dynamic parameter identification method for industrial robots that adaptively selects optimal experimental data samples. The approach addresses the challenge of maintaining motion control accuracy when end-effector payloads change frequently in precision manufacturing tasks such as drilling and assembly.
Key takeaways
- Method uses Newton-Euler recursive modeling to derive payload dynamics by subtracting unloaded from loaded robot models, avoiding the need to identify base robot parameters.
- Only the first three joints are used for identification, reducing data coupling and improving accuracy compared to full six-joint approaches.
- Fourier series excitation trajectories are optimized using Hadamard inequality, and an adaptive sample selection strategy filters collected data to form an optimal sample set.
- Weighted least squares estimation yields payload parameters with high accuracy across different load scenarios, while reducing data volume for more efficient online identification.
Experimental validation showed the method maintains high identification precision under various loads, outperforming traditional techniques. By optimizing sample selection, it reduces the amount of data required, offering a practical pathway for real-time payload parameter identification in industrial robots.
Source: 《机器人》期刊 (robot.sia.cn) · Published 2026-01-13 · “基于最优试验数据样本自适应选择的工业机器人负载动力学参数辨识方法”