Robot Manipulation Skills Transfer for Sim-to-Real in Unstructured Environments

• DOI: 10.3390/electronics12020411
• Publication Date: 2023-01-13T07:29:57Z
• Authors: Zikang Yin; Chao Ye; Hao An; Weiyang Lin; Zhifeng Wang
• Keywords: Social robot; Arm solution; Impedance Control; Personal robot

Robot force control that needs to be customized for the robot structure in unstructured environments with difficult-to-tune parameters guarantees robots’ compliance and safe human–robot interaction an increasingly expanding work environment. Although reinforcement learning provides a new idea adaptive adjustment of these parameters, policy often trained from scratch when used robotics, even same task. This paper proposes episodic Natural Actor-Critic algorithm action limits improve admittance transfer motor skills between robots. The motion learned by simple simulated robots can applied complex real robots, reducing difficulty training time consumption. ensures realizability mobility robot’s all directions. At time, builds up environment model realizes impedance during movement. In typical contact tasks, are simulation reality perform performance each task is similar environment, which verifies method’s effectiveness.