Adaptive model-free disturbance rejection for continuum robots

Abstract

This paper presents two model-free control strategies for the rejection of unknown disturbances in continuum robots. The strategies utilize a neural network-based approximation technique to estimate the uncertain Jacobian matrix using position measurements. The first strategy is designed for periodic disturbances and employs an adaptive model-free controller in conjunction with an adaptive disturbance observer. The second strategy is designed for robustness against arbitrary disturbances and employs time-varying input and update law gains that grow monotonically, resulting in the achievement of asymptotic, exponential, and prescribed-time reference trajectory tracking. The notion of fixed-time stabilization in prescribed time is particularly noteworthy, as it allows for the predefinition of a terminal time, independent of initial conditions and system parameters. A formal stability analysis is presented for each strategy, and the strategies are both tested experimentally with a concentric tube robot subject to unknown disturbances.