A disturbance rejection control for urban air mobility using artificial sensor-based Gaussian process regression

Abstract

This paper presents a learning-based disturbance rejection control strategy for Urban Air Mobility (UAM) with vertical take-off and landing capability, which is subject to uncertainties in system parameters. The two primary sources of uncertainty during UAM operation, specifically moment of inertia uncertainty and center of gravity variation, are thoroughly analyzed as they negatively impact control performance. Building upon the analysis outcomes, a novel adaptive scheme is proposed that employs the modeling capabilities of Gaussian process regression for online learning to estimate model uncertainties. To ensure the collection of high-quality training data without relying on state derivatives, a nonlinear disturbance observer is employed as an artificial sensor. The suggested control algorithm is formulated by integrating Gaussian process regression with a baseline control derived using the feedback linearization control technique. Theoretical analysis grounded in the Lyapunov theorem reveals that the tracking error of the closed-loop system is semi-globally uniformly and ultimately bounded. Numerical simulations are conducted to validate the effectiveness of the proposed approach. The results obtained confirm that the proposed method can achieve superior tracking performance, even in the presence of model uncertainties and time-varying disturbances, surpassing existing approaches.