Research On Decision-making Method For Unmanned Surface Vehicle Collision Avoidance Considering Maritime Collision Avoidance Rules

In order to meet the increasing demand for resources,countries around the world are competing to carry out marine science research and ocean resource development,as the ocean contains abundant marine resources.USVs,as the main tool for exploring complex unknown marine environments,are widely used in marine development and military activities,with enormous economic and military value,and have attracted the attention of many experts and scholars.The navigation of USVs involves various aspects such as environmental perception,navigation path planning,autonomous collision avoidance decision-making,and motion control.Among them,autonomous collision avoidance decision-making is the core issue of safe,efficient,and energy-saving navigation of USVs on the sea.Therefore,the study of collision avoidance methods for USVs has significant significance.In addition,marine collision accidents are related to human decision-making errors in complying with COLREGs.Therefore,it is also necessary to consider COLREGs as an important factor in the collision avoidance decision-making of USVs.The main research content of this thesis include:(1)With respect to USV collision avoidance,this thesis clarifies the collision avoidance decision-making process of USVs based on collision avoidance rules.Firstly,a collision risk assessment module based on DCPA and TCPA is established.In view of the situation where the target vessel does not comply with COLREGs,a multi-level collision risk assessment method is proposed,considering the type of ship and navigation environment,setting different thresholds,and comparing DCPA and TCPA parameters to determine the collision risk level of USVs.Then,a COLREGs rule selection module is established,which analyzes the USV encounter situation according to COLREGs,and clarifies the collision avoidance responsibility and collision avoidance operations to be taken under different encounter situations.Finally,a collision avoidance path planning module is established.In order to meet the real-time requirements of USV collision avoidance,a sub-waypoint is generated to plan the collision avoidance path.(2)A USV collision avoidance path planning method based on a multi-objective jellyfish search algorithm is proposed.Firstly,the USV collision avoidance problem is solved as a multi-objective optimization problem.An adapted function based on collision avoidance path safety,smoothness,and economy is used to evaluate the USV collision avoidance problem,considering safety,COLREGs,USV maneuverability,and other requirements,and constructing constraints for the collision avoidance problem.Secondly,a multi-objective jellyfish search algorithm is used to find the Pareto optimal solution to the multi-objective optimization problem.Finally,numerical simulations of four encounter situations and complex encounter situations are conducted for the proposed collision avoidance path planning method.The simulation results show that the proposed method can effectively achieve safe collision avoidance between two ships and in complex encounter situations.Semi-physical simulation platform verification tests further demonstrate the effectiveness of the proposed method in safe collision avoidance in complex encounter situations.At the same time,the optimization performance of the multi-objective particle swarm algorithm and the multi-objective jellyfish search algorithm is compared,and the results show that the collision avoidance decision-making method based on the multi-objective jellyfish search algorithm has higher efficiency and optimization accuracy in searching for the optimal collision avoidance path in complex encounter situations.(3)A new multi-objective constraint avoidance path planning model was developed to enable safe navigation in restricted waters using intelligent optimization algorithms,by adding new constraints.The multi-objective jellyfish search algorithm was improved by introducing a two-stage continuous mutation strategy and an adaptive factor.By performing mutation operations on both the local and global optimal positions obtained from each update of the jellyfish population,the algorithm’s convergence speed and optimization accuracy were improved while avoiding local optima.A turning path smoothing method was proposed,applying a turning radius constraint to the ship,to obtain a smooth avoidance path that matches the actual turning behavior of the vessel.Subsequently,the effectiveness of the proposed path planning method based on the improved multi-objective jellyfish search algorithm was verified through numerical simulation tests under four different encounter situations,different path preferences,and restricted water environments.Compared to the multi-objective particle swarm algorithm and the original multi-objective jellyfish search algorithm,it had faster search efficiency for optimal paths and higher optimization accuracy.