Abstract
Autonomous operation is a crucial feature for hydraulic manipulators to perform specialized tasks, and the effectiveness of online autonomous operation relies on the accuracy of the dynamic model. To tackle the issue of low accuracy of the dynamic model when a hydraulic manipulator grasps an unknown payload, a method of online model correction is proposed in this paper. In comparison to other online identification methods that simplify the inertia parameters of payload, an excitation trajectory based on physical feasibility is firstly designed to correct the dynamic model with non-payload under the offline identification. A forgetting factor is introduced into the recursive least squares (RLS) algorithm to further improve the effectiveness of the identification since it can continuously correct the dynamic model with the iterative formula. Through accurately identifying the complete set of inertia parameters for payload, the dynamic model is corrected in real-time. The proposed method is experimentally validated under the conditions of variable payload, and the accuracy of the corrected dynamic model is improved.
