Obstacle Avoidance during Teleoperation by Model Predictive Control with Time-varying Delay

摘要

This paper proposes a method for autonomously replanning the optimal trajectory of a teleoperated mobile robot to avoid obstacles during teleoperation with time-varying delay. Model predictive control (MPC) is applied to compensate for the time-varying delay to ensure the mobile robot autonomously avoids obstacles. An operator operates the robot while viewing the predicted model. The state of the predicted model is sequentially sent to the teleoperation system by the local system including the predicted model. This earlier received predicted state is used as a reference trajectory for MPC, and linear interpolation is performed on the lost data of the reference trajectory. When an obstacle is encountered, the path to avoid it is replanned. The constraints for obstacle avoidance, including the cost of the distance between the obstacle and the mobile robot are relaxed, making obstacle avoidance control robust to modeling errors. The mobile robot then follows a reference trajectory while maintaining the optimal distance to the obstacle. Simulations and experiments are conducted to evaluate the performance of the proposed method in comparison with conventional methods.