Research on Door Opening Operation of Mobile Robotic Arm Based on Reinforcement Learning

The traditional robotic arm control method has strong dependence on the application scenario. To improve reliability of mobile when scene is disturbed, this paper proposes a based an improved proximal policy optimization algorithm. This study researches arms for opening doors. At first, door handle position obtained through image-recognition YOLOv5. Second, simulation platform CoppeliaSim used to realize interaction between and environment. Third, strategy reward function designed train applied opening-door task in real experimental results show that proposed can accelerate convergence training process. Besides, our effectively reduce jitter stability control.