Semi-autonomous Telemanipulation
Teleoperation solves tasks in which a human operator can remotely manipulate objects using the teleoperated robot, essential in industrial inspection and repair, space exploration, search and rescue, and assistive living robotics. The goals of my research are: 1) investigating the methodology to convey the human command to the robot with intuitive motion mapping through autoencoder, 2) enabling robots to semi-autonomously provide effective assistance yet still accommodating the operator’s commands for telemanipulation, which can reduce the human’s burden to overcome the control ambiguity and structural discrepancy. Then transferring the knowledge to assist human operation between different robots to generalize the semi-autonomous telemanipulation method. 3) developing a transferability-based chain motion mapping method to scale up the teleoperation scheme to different robots. 4) proposing a hierarchical physics-informed semi-autonomous reinforcement learning framework with multi-stage training strategies to ensure grasp stability while following human dynamic motion.
Learn and Transfer Knowledge of Preferred Assistance Strategies in Semi-Autonomous Telemanipulation
Lingfeng Tao, Michael Bowman, Xu Zhou, Jiucai Zhang & Xiaoli Zhang
Journal of Intelligent & Robotic Systems
Enabling robots to provide effective assistance yet still accommodating the operator’s commands for the telemanipulation of an object is very challenging because robot assistance is not always intuitive for human operators and human behaviors and preferences are sometimes ambiguous for the robot to interpret. Due to the difference in hand structures, some motion assistance from the robot may surprise the operator with counter-intuitive movements, which could introduce more burden to the human to correct the actions and/or reduce the operator’s sense of system control. To address these problems, we developed a novel preference-aware assistance knowledge-learning approach. An assistance preference model learns what assistance is preferred by a human, and a stage-wise model updating method ensures learning stability while dealing with the ambiguity of human preference data. Such preference-aware assistance knowledge enables a teleoperated robot hand to provide more active yet preferred assistance toward manipulation success. We also developed knowledge transfer methods to transfer the preference knowledge across different robot hand structures to avoid extensive robot-specific training. Experiments were conducted to telemanipulate a 3-finger hand and a 2-finger hand, respectively, to use, move, and hand over a cup. Results demonstrated that the methods enabled the robots to learn the preference knowledge effectively and allowed knowledge transfer between robots with less training effort.
