Advances in Observer-Based Line-of-Sight Guidance for Autonomous Vehicles: A Survey

Autonomous vehicles (AVs) face numerous challenges when performing path-following tasks in complex environments such as oceans, rivers, and lakes. In recent years, Line-of-Sight (LOS) guidance has gained widespread attention as an efficient guidance method. This paper reviews the latest advancements in observer-based LOS guidance technology for AVs. First, the main LOS guidance method and its three variants are thoroughly analyzed. Next, the applications of four traditional observers and neural network-based observers in LOS guidance are discussed. These observers enhance the systems robustness and path-following accuracy by enabling real-time estimation of external disturbances and system uncertainties. Finally, four key areas for future research are proposed, including multi-observer fusion, distributed cooperative observation, hybrid observers combining online and offline learning, and the optimization of lightweight neural network-based observers.