Feedback linearizing control for moving-base line-of-sight tracking systems via adaptive spherical target estimation

This paper investigates the maneuvering target tracking problem for moving-base line-of-sight (LOS) tracking systems under conditions of unknown observation noise. First, we establish a comprehensive system model that simultaneously considers the LOS pointing model and the gimbal dynamics model in the LOS tracking system. Then, an advanced tracking error feedback control system based on feedback linearization principles is designed, effectively transforming target maneuvers and unmatched base disturbances into matched disturbances that are directly compensated through a disturbance observer, achieving superior control performance compared to a traditional gyroscope-feedback-based system. Furthermore, the time derivative of tracking error obtained by conventional command estimation methods is vulnerable to the deleterious impact of base motion. To overcome this limitation, we propose a spherical target estimation (STE) approach, which estimates the target's motion on a spherical surface in the world frame. This eliminates the coupling effect of base motion in the time derivative of tracking error and does not require target distance or gimbal attitude sensors. In addition, to tackle the changes in the properties of measurement noise caused by target detection and camera characteristics, a variational Bayesian Kalman adaptive filter (VB-AKF) is introduced. This method effectively reduces target motion estimation errors caused by measurement noise variance mismatch, thereby enhancing tracking performance. Simulation results in the Webots robotic simulator demonstrate the superior performance of the proposed methods.