This paper focuses on collision avoidance (CA) for ships under multiple uncertainties. A ship CA framework is designed combining robust motion control of the Own Ship and probabilistic prediction of the Target Ships’ behavior. A motion control method based on the Tube-based Model Predictive Control (MPC) is designed to achieve robust trajectory tracking, considering uncertainties about ship motion and external disturbances. A high-precision probabilistic trajectory prediction method based on GPR with the incremental theory is proposed to describe the uncertain behavior of the TSs. The artificial potential field (APF) method is introduced to deal with the CA constraints in Tube-based MPC, effectively reducing computational complexity. Simulation experiments with different degrees of uncertainty demonstrate the effectiveness of the proposed framework for ship CA.

中文翻译：

---

多重不确定性下船舶避碰的鲁棒模型预测控制


本文重点研究多重不确定性下船舶的防撞（CA）。船舶CA框架的设计结合了本船的鲁棒运动控制和目标船行为的概率预测。考虑到船舶运动和外部干扰的不确定性，设计了一种基于管基模型预测控制（MPC）的运动控制方法，以实现鲁棒的轨迹跟踪。提出了一种基于增量理论的基于探地雷达的高精度概率轨迹预测方法来描述TS的不确定行为。引入人工势场（APF）方法来处理Tube-based MPC中的CA约束，有效降低计算复杂度。不同程度不确定性的仿真实验证明了所提出的船舶CA框架的有效性。