Abstract
Friction nonlinearity near zero velocity causes substantial tracking errors during joint motion reversals. For hybrid robots, this phenomenon is further influenced by joint acceleration and robot configuration, unique characteristics of hybrid robots that can degrade the performance of traditional friction compensation methods. This paper presents a novel multi-pulse friction compensation strategy that can adapt to joint acceleration and configuration variations in hybrid robots. Bayesian Optimization is employed to automatically tune all compensation parameters. By analyzing experimental data, a potential relationship between compensation parameters and joint acceleration is explored, leading to a concise and effective method for estimating optimal parameters based on joint acceleration. In addition, the basic idea of cluster analysis is combined with a limited number of experiments to achieve online parameter-to-configuration matching. Experimental results on TriMule-200 hybrid robot demonstrate the outstanding performance of this strategy in suppressing tracking errors during velocity reversals, as well as its robustness to joint acceleration and robot configuration variations.
Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
