Motion prediction for leader manipulator of teleoperation system with large time delay based on inverse optimal...

This article presents a novel motion prediction method for the leader manipulator of a teleoperation system with large time delay. First, an impedance controller is designed for the leader device to obtain a linear human-in-the-loop dynamic model based on a sliding mode disturbance observer. Then, a new inverse optimal control (IOC) approach is applied to identify human operational intentions through offline demonstrations, which only require the position and velocity measurements of the robot end-effector. By using the identification results, a new cascade predictor is designed, which consists of a continuous-discrete time observer and several sub-predictors in a chain, such that the real-time motion of the leader manipulator can be estimated by using the discrete delayed state measurements. The convergence of prediction errors under arbitrarily large time delays is demonstrated via the Lyapunov-Krasovskii method. Finally, the effectiveness of the proposed predictor is verified through simulations and experiments.