Path Following Controller for Snake Robots Based on Data-driven MPC and Extended State Observer

Abstract

This paper investigates the straight path following control problem for snake robots without any prior model knowledge. A data-driven model predictive control (MPC) scheme is designed based on a data-driven Koopman model and an extended state observer (ESO). Firstly, this paper establishes a data-driven model for heading control of the snake robot based on Koopman operator theory. The Koopman operator enables global linear representations of nonlinear dynamical systems in a high-dimensional space. Then, a MPC is designed based on the high-dimensional linear Koopman model, which can solve the optimal control input that satisfies the constraints in real-time. Moreover, the model uncertainty is estimated by ESO and compensated in the controller, which improves the robustness of the controller. Finally, the effectiveness and superiority of the proposed controller are verified by simulations.