VRMP: towards efficient human-robot cooperative teleoperation via digital twin-powered motion hybrid mapping and prediction

Human-robot cooperative teleoperation can assist human in controlling robots to execute intricate tasks in hazardous, extreme, or disaster scenarios. This is a current research focus in the field of robotics. However, existing teleoperation research faces challenges related to complex systems and high proficiency requirements for operators, resulting in low efficiency in human-robot collaboration. Considering the above situation, we propose a human-robot collaborative teleoperation framework based on digital twin and trajectory prediction, called VRMP. Firstly, we establish the digital space for the remote robot operation scene and create the virtual scene through virtual-real mapping. A universal hybrid mapping strategy is then proposed to enhance human-robot collaboration. Finally, a method for predicting manipulator trajectory driven by digital space is proposed. The experimental results of real human-robot cooperative operation show that the proposed system reduces the time for obstacle avoidance and grasping tasks by 22% compared to traditional teleoperation. Additionally, the success rate is increased by 16%