Improved interfered fluid dynamic system based LOS for USV path following with obstacle avoidance capability

In order to enhance the consistency of both path following and obstacle avoidance, as well as to improve the autonomous performance of intelligent vessels, this paper presents a novel guidance-and-control framework, including an IIFDS-LOS, a rule-based adaptive MPC controller, and an ensemble learning-based compensation strategy. The proposed guidance law not only calculates the expected heading angle during the path following process, but also directs USVs in steering clear of obstacles in compliance with the COLREGs. Compared to other derivative methods, the IIFDS-LOS improves the rationality of dodging behavior, addresses the limitation of delayed avoiding actions, and enables USVs to achieve avoidance with smaller steering angles. Subsequently, a rule-based steering controller is implemented, taking into account various constraints during both the tracking and avoiding stages. Besides, an ensemble NN-based compensator is introduced to calibrate discrepancies between the actual model and the nominal model, with the aim of enhancing model accuracy and control effect. Finally, the practicality and superiority of the framework are validated through numerical simulations and physical experiments. The synthesis results demonstrate that the proposed scheme exhibits superior robustness and stability in path following tasks under bounded interference, while also facilitating timely avoidance maneuvers when encountering obstacles.