Study On Ship Autonomous Collision Avoidance In Complex Waters

With the rapid development of the maritime transportation industry,the navigation of ships in estuaries,ports,Straits,and other waters becomes continuously intense.The traffic environment is increasingly complex,and the behavior of ships is changeable,which is easy to lead to a dangerous encounter situation,and ship collision accidents occur from time to time.Autonomous navigation is an important means to ensure ship safety.Autonomous collision avoidance decision-making is the core of autonomous navigation.Given the gap between the decision-making and navigation practice and based on the key factors of collision avoidance decision-making oriented to navigation practice,this research proposes an autonomous collision avoidance decision-making method in complex waters in line with navigation practice based on collision avoidance mechanism,which fully considers the constraints of ship maneuverability,target ship motion uncertainty,collision avoidance rules,and good seamanship.Such a method could provide decision support for ship navigation safety,and a useful reference for intelligent navigation in the future.The main contents of the research are as follows:(1)The key factors of collision avoidance decision-making in complex waters are comprehensively analyzed.Through bibliometric and questionnaire surveys,the commonalities and differences between the key factors of collision avoidance decisionmaking of ships in theoretical research and navigation practice are explored,and the key factors of collision avoidance decision-making of ships in complex waters are deeply differentiated,so as to provide an important reference for the modeling of autonomous collision avoidance decision-making of ships in complex waters.(2)A novel autonomous collision avoidance decision-making framework for navigation in complicated waterways is proposed.Based on the key factors of collision avoidance decision-making in complex waters,this research established a collisionavoidance decision-making framework for navigation practice following the kay principle of “target motion prediction-conflict detection-conflict resolution”,which provides a theoretical basis for subsequent collision avoidance decision-making modeling;(3)The prediction model of target ship motion based on data-driven is put forward.Aiming at the influence of the uncertainty of the TS motion on the ship’s collision avoidance decision,the route model containing the traffic mode and the ship navigation law is extracted through data mining.The target ship motion uncertainty is described by the probability model,and the target ship trajectory prediction model based on the Gaussian process is established,which provides the basis for the modeling of collision risk and collision avoidance decision-making.(4)The collision risk model based on the probability velocity obstacle method is established.To overcome the challenges from multiple influencing factors of the collision risk modeling in the complicated waterways,this research utilized probabilistic velocity obstacle as the fundamental framework to integrate the COLREGs,good seamanship,ship maneuverability,target ship motion uncertainty into the risk measurement in the velocity domain of the own ship,which improves the accuracy of the risk calculation results,reflects the ship encounter process more realistically and provides a good foundation for the collision avoidance decisionmaking.(5)The autonomous ship collision avoidance decision model considering the whole process of navigation practice is put forward.This paper constructs the collision avoidance decision-making method under the influence of multiple factors based on the collision avoidance mechanism with the consideration of the target trajectory prediction and collision risk quantification results.Firstly,based on the nonlinear velocity obstacle method,a rolling optimization ship avoidance model is established by integrating the factors such as COLREGs and good seamanship,ship maneuverability,and uncertainty of target ship motion in the velocity space.Then,on the premise of ensuring safety,combined with the improved Line Of Sight(LOS)method route correcting model,this paper constructs a collision-avoidance decisionmaking model in complex waters for navigation practice and weighs the economy and safety,so as to realize the autonomous collision avoidance decision-making of ships in complex waters.(6)A test platform base on navigational big data is established to validate the developed collision-avoidance decisions.Based on AIS and other navigation big data platforms,the simulation of the practical navigation environment with higher fidelity is realized with the modifications to Open CPN software,and the collision avoidance decision-making model proposed in this paper is embedded in the platform.The collision avoidance decision-making experiment is carried out with real encounter scenarios,which verifies the feasibility and reliability of the method proposed in this study in the real sea environment.Considering the achievement and methods proposed in the research,the novelties of the research can be concluded as follows:(1)Based on the systematic analysis of practical knowledge and experience of navigation,based on the key factors and requirements of intelligent collision avoidance decision-making in the future,a decision-making method system for autonomous collision avoidance in complex waters is proposed,which conforms to the navigation practice,and provides a useful reference for the modeling of collision avoidance decision-making;(2)Based on the probability velocity obstacle method,a novel collision risk model is established by considering target ship motion uncertainty,ship maneuverability,and COLREGs,which is suitable for the complicated navigation waterways with multiple encounters and could estimate the collision risk from the perspective of navigation practices and good seamanship.(3)The ship autonomous collision avoidance decision model in line with navigation practice is established for complicated waters,following the fundamental principle of “target ship motion prediction-conflict detection-conflict resolution”.By integrating the COLREGs and good seamanship into the decision-making process,the results of the proposed method are applicable for both the conventional manned ship and autonomous ship.In conclusion,this research takes the collision avoidance decision-making in complicated waters as the main research subject and focused on the key challenges and the whole process of collision avoidance decision-making to establish a novel collision avoidance decision-making model,which takes the navigational practices,COLREGs,ship maneuverability,motion uncertainty of target ships and good seamanship,etc.into consideration using the velocity obstacle methods.The methodological framework for autonomous collision avoidance decision-making is also innovated.The research has promising potential for practical application,which is beneficial for the development of an intelligent and autonomous maritime traffic system.