This paper investigates friction modeling and compensation for haptic master manipulator used in robot-assisted minimally invasive surgical system. Friction modeling and compensation is based on deep Gaussian process (DGP) and it does not require the utilization of explicit friction models. Therefore, the proposed friction modeling and compensation algorithm can circumvent the drawbacks associated with model-based methods. Through the adoption of implicit posterior variational inference, the proposed algorithm can accurately predict and compensate friction even when there exists parametric uncertainty in the dynamics of the haptic master manipulator. The effectiveness and feasibility of the proposed approach is validated experimentally through the robot-assisted minimally invasive surgical system. Experimental results demonstrate that the proposed method outperforms several existing alternatives in representing and compensating friction effects.

本文研究了用于机器人辅助微创手术系统的触觉主机械手的摩擦建模和补偿。摩擦建模和补偿基于深高斯过程 (DGP)，不需要使用显式摩擦模型。因此，所提出的摩擦建模和补偿算法可以避免与基于模型的方法相关的缺点。通过采用隐式后验变分推理，即使在触觉主操纵器的动力学中存在参数不确定性的情况下，所提出的算法也可以准确地预测和补偿摩擦。通过机器人辅助微创手术系统实验验证了所提出方法的有效性和可行性。