A Computationally Effective Multi-Output Recursive Gaussian Process-based Path-following Guidance for Unmanned Aerial Vehicle

Abstract

In this paper, we present a path-following guidance algorithm for unmanned aerial vehicles, leveraging Gaussian processes (GPs) to achieve precise tracking performance even in the presence of external disturbances. The baseline guidance algorithm is developed by integrating guidance kinematics and optimal error dynamics. To model disturbances along each axis, GPs are employed. To address computational constraints, we introduce a novel joint recursive GP approach that combines the strengths of recursive GPs and multi-target-based GPs. Through numerical simulations, we examine the estimation performance of the proposed algorithm.