Redundant robot control with learning from expert demonstrations
Jorge Ramírez, Wen Yu · 2022 IEEE Symposium Series on Computational Intelligence (SSCI) · 2022
This paper proposes a biased exploration based reinforcement learning, which uses expert experiences to avoid the exploration of all states. The method is applied to control redundant robots with expert experiences. A 7-degree-of-freedom robot manipulator is used in experiments. The results show that expert demonstrations based robot control works well.