Inverse-Free Tracking Control of Continuum Robots with Unknown Models Based on Gradient Neural Networks

摘要

The inherent dexterity and compliance of continuum robots have endeared them to numerous researchers. However, the control of continuum robots remains a complicated problem worth studying as a result of their intricate structures. In this work, we present a scheme based on gradient neural networks (GNN) for the control of continuum robots with unknown models. The proposed scheme is composed of two GNN models, one of which is employed for the solution of inverse kinematics problem, and the other is used to estimate the Jacobian matrix of continuum robots. This design allows us to rely only on user-defined input and sensory output to achieve the tracking control of continuum robots, without knowing their models and internal structures. The convergence of the proposed scheme is proven by theoretical analysis. Finally, the feasibility and merits of the proposed scheme are revealed by simulation studies, including performance analysis and comparisons.