Learning-Based Predictive Path Following Control for Nonlinear Systems Under Uncertain Disturbances

Accurate path following is challenging for autonomous robots operating in uncertain environments. Adaptive and predictive control strategies are crucial a nonlinear robotic system to achieve high-performance control. In this paper, we propose novel learning-based scheme that couples high-level model controller (MPFC) with low-level feedback linearization (LB-FBLC) systems under disturbances. The LB-FBLC utilizes Gaussian Processes learn the environmental disturbances online tracks reference state accurately probabilistic stability guarantee. Meanwhile, MPFC exploits linearized augmented virtual linear dynamics optimize evolution of targets, provides states controls LB-FBLC. Simulation results illustrate effectiveness proposed strategy on quadrotor task unknown wind