Adaptive Robust Control for Bimanual Cooperative Contact Teleoperation with Relative Jacobian Matrix

Due to the time delay in bilateral teleoperation, human reaction lags behind changes in the object state and the operating environment, which cannot solve unexpected accidents. An adaptive robust control strategy is proposed in this paper to address this problem. Inspired by human operational habits, two slaves are first divided into the dominant slave and the non-dominant slave with different functions. The non-dominant slave helps the dominant slave in humanoid style, as measured by relative Jacobian matrix. Then, the slaves’ reactions are divided into three stages: initial stage, adjustment stage, and recovery stage. In the second stage, the non-dominant slave helps the dominant slave according to its current state, and then they both track the master’s state in the third stage. The controllers of the two slaves are also designed differently. The dominant slave follows the dominant hand’s motions. Thus, an adaptive method is used based on RBF-neural networks to suppress uncertain system dynamics and forces. The non-dominant slave controller is designed in three stages consisting of three parts: the relative term, designed for collaborative motions; the shared term for sharing control signals from the dominant side; and the independent term for processing local disturbance. Finally, numerical experiments show the effectiveness of the proposed teleoperation architecture.