In context
In 2023, dual-arm collaborative robots were gaining traction in manufacturing for tasks like welding, assembly, and polishing, but their effectiveness hinged on cooperative positioning accuracy. Geometric and non-geometric errors—from manufacturing tolerances, joint compliance, and gear backlash—degraded performance, yet most calibration methods focused on single-arm accuracy or ignored non-geometric errors in dual-arm systems. This paper addressed that gap with a hybrid calibration approach.
What was reported
Researchers at Hebei University of Technology proposed a two-step kinematic calibration method combining parametric and non-parametric models for a dual-arm system comprising UR10 and UR5 robots. First, they built a modified Denavit-Hartenberg (MDH) model and identified geometric parameter errors using an iterative least squares method after removing coupling parameters. Second, they used a backpropagation neural network to predict and compensate non-geometric errors of each arm, overcoming the limitation of traditional methods that only worked within the calibration coordinate system.
For the dual-arm system, they identified the base frame transformation matrix parameters based on distance errors and compensated both geometric and non-geometric errors. Experiments showed the method reduced average positioning errors of UR10 and UR5 to 0.1709 mm and 0.0509 mm, respectively. The average cooperative positioning error dropped to 0.1676 mm—a 27.7% improvement over a parametric-model-only approach.
Why it mattered
This work demonstrated that incorporating non-geometric error compensation into dual-arm calibration can significantly enhance cooperative precision, which is critical for high-accuracy manufacturing tasks. The method offers a practical framework for system integrators to improve dual-arm robot performance without hardware changes, potentially expanding their use in precision assembly and finishing operations.
"The average cooperative positioning error of the dual arms is reduced to 0.1676 mm, and the accuracy of the proposed method is improved by 27.7% compared with the method based on the parametric model."
Source: 《机器人》期刊 (robot.sia.cn) · Published 2023-05-09 · “基于参数与非参数模型结合的双臂机器人协作定位精度提升方法”