Robust Nonlinear Neural Network-Based Control of a Haptic Interaction with an Admittance Type Virtual Environment

For simulating the surgical procedures in a virtual environment, it is necessary to propose a suitable virtual environment model which can reflect the real physical tool-tissue interaction behavior and a proper human user interface appropriate dynamic model. In this study, a linear Kelvin-Voigt and a nonlinear Hunt-Crossley models have been utilized to describe human hand dynamics during interaction with a haptic interface. Using the Lyapunov stability criteria, an adaptive neural network based controller being designed for guaranteeing the stability of the entire system, considering a nonlinear model for the environment, an inertia for a virtual tool, a constant time delay for data transforming from environment to the operator's side, and external disturbances. Results show that even if both hand models are stable in the sense of Lyapunov stability criteria, nevertheless, the nonlinear Hunt-Crossley model is less sensitive to the hand stiffness escalation and is more transparent than the linear Kelvin-Voigt model, especially in the high-frequency input forces.