It is crucial to develop a COLREGs-compliant intelligent collision avoidance system for the safety of unmanned ships during navigation. This paper proposes a collision avoidance decision approach based on the deep reinforcement learning method. A modified collision avoidance framework is developed that takes into consideration the characteristics of different encounter scenarios. Hierarchical reward functions are established to assign reward values to constrain the behavior of the agent. The collision avoidance actions of the agent under different encounter situations are evaluated on the basis of the COLREGs to ensure ship safety and compliance during navigation. The deep Q network algorithm is introduced to train the proposed collision avoidance decision framework, while various simulation experiments are performed to validate the developed collision avoidance model. Results indicate that the proposed method can effectively perform tasks that help ships avoid collisions in different encounter scenarios. The proposed approach is a novel attempt for intelligent collision avoidance decisions of unmanned ships.

It is vital to analyze ship collision risk for preventing collisions and improving safety at sea. This paper takes Ningbo-Zhoushan Port, a typical complex navigable water, as the research object. Firstly, a probabilistic conflict detection method based on an AIS data-driven dynamic ship domain model is proposed to achieve effective ship conflict detection under uncertain environments. Then, a ship group identification method is proposed, which can extract the ship groups with conflict correlation and space compactness. Finally, according to the characteristics of ship traffic in complex navigable waters, the dynamic calculation of ship collision risk is carried out from individual, regional, and local multi-scale perspectives. The experimental results show that the proposed method can detect the collision risk in a timely, reliable, and effective manner under complex dynamic conditions. As such, they provide valuable insights into ship collision risk prediction and the development of risk mitigation measures.

It is vital to analyze ship collision risk for preventing collisions and improving safety at sea. This paper takes Ningbo-Zhoushan Port, a typical complex navigable water, as the research object, propose a probabilistic conflict detection method to estimate the potential collision risk by using a dynamic domain model driven by AIS data. Combined with the algorithm of fast modularity optimization and spectral clustering, a ship group extraction method from the perspective of water navigation management was proposed. Aiming at ship traffic characteristics in complex navigable waters, conduct dynamic calculation of ship collision risk from individual, regional and local multi-scale perspectives. The results show that the proposed method can detect the collision risk in a timely, reliable and effective manner under complex dynamic conditions. As such, they provide valuable insights for ship collision risk prediction and the development of risk mitigation measures.

Assessing ship conflicts in complex inland waterways is challenging because constrained navigation environments and heterogeneous ship interactions limit the applicability of generic safety thresholds. This study proposes a behavior-informed conflict assessment method using AIS data. AIS trajectories are first processed to detect candidate maneuvers from heading and speed variations. Geometry-induced turning behavior is then filtered using waterway spatial constraints, and the remaining maneuvers are associated with crossing and overtaking encounters to identify avoidance-onset samples. These samples are used to optimize the parameters of encounter-specific ship-domain models, thereby linking domain construction directly to observed navigator behavior. Empirical analysis of AIS data from the Taicang reach of the Yangtze River yielded 10,256 valid ship trajectories. Passenger vessels and tugboats were excluded because their operating patterns differ from those of merchant vessels. Based on the optimized domains, ship conflicts were identified and quantified through domain overlap. Spatial analysis identified two descriptive concentration regions associated with crossing conflicts, while overtaking conflicts showed a more continuous distribution along the main channel. These patterns provide interpretable evidence for targeted inland-waterway safety management.