Tremor Attenuation For Robot Teleoperation By A Broad Learning System-Based Approach

Abstract

It is a meaningful yet challenging task to achieve the synchronous of a teleoperated robotic manipulator and a human operator, for which one of the key problems to be overcome is the elimination of the physiological tremor from human hands. To address the problem, a tremor attenuation filter based on the broad learning system is established. This filter shows universal approximation properties and fast learning speed. Meanwhile, different trajectories have different network optimal parameters. Hence, incremental learning algorithms are combined because they only need to learn new information and can achieve fast-updating to obtain the desired performances. Simulation results confirm the effectiveness of the proposed method.