Neural Network Control of a Robot Interacting With an Uncertain Viscoelastic Environment

A continuous controller is developed for a robot that moves in free space, undergoes a collision with a viscoelastic environment, and then regulates the new coupled dynamic system to a desired setpoint. Since the model for the viscoelastic surface contains uncertainties that do not satisfy the linear-in-the-parameters assumption, the model is approximated by a neural network feedforward term, which is combined with a continuous feedback term to guarantee uniformly ultimately bounded regulation of the system despite parametric uncertainties in the robot and the viscoelastic environment. Experimental results of a two-link robot interacting with a human tissue phantom demonstrate the performance of the proposed controller.