Research On USV Collision Avoidance Decision Method Based On Multiple Objects

Different from the ship collision avoidance decision-making system,unmanned collision avoidance decision-making system has no person participate,so need higher technology.This subject provides collision decision-making system for the unmanned vessel experiment in Shanghai.The system accepts the plan route from the control console and receives the sensor information,then sends the collision avoidance decision to the navigation controller,so that the unmanned vessel can avoid the static obstacles and dynamic obstacles when navigating according to plan.The first step of the collision avoidance process is dealing with the information to the parameters that system needs,includes motion parameters,segment judgment coordinate conversion and so on.Secondly,the paper combines the the experience of experts to establish the minimum security meeting area model,and then assesses risk with these parameters.If there is risk,the algorithm will calculate a safe path and output to the controller next.The paper realizes the collision avoidance decision-making system with artificial collision field method and geometric collision avoidance method respectively.And evaluation index is established to assess the action.For the artificial potential field collision avoidance decision system,the paper establishes the artificial potential field,solves the local minimum problem,then builds the static environment models with geometric modeling,and establishes dynamic obstacle models with considering the safety.Finally,the paper proposes a path point filter model to solve the problem that artificial potential field outputs too much path points to meet the controller requirements.For the geometric avoidance decision-making system,the paper introduces the basic concepts and geometric collision avoidance principles briefly,builds the model of avoiding high speed vessel and the model of avoiding low speed vessel.Then discusses the meeting situation and the action range of maritime rules,combines the geometric avoidance model with the maritime rules,establishes the safe heading model and recovery timing model.Finally,the paper proposes three solutions for geometrically avoiding multiple vessels,and analyze every solutions' performance.The simulation results of artificial potential field method show that the method can avoid dynamic and static obstacles,but the scores of avoidance action are not high,and the actions don't comply with the maritime rules.The experimental results show that the problem of controller delay need to be considered.The simulation results of geometric collision avoidance system show that the method can avoid dynamic obstacles better than artificial potential field method.In summary,the artificial potential field method is better to avoid static obstacles than geometric avoidance,and worse to avoid dynamic obstacles than geometric avoidance.The geometric avoidance can combine with maritime rules,the action and range of avoidance can be limited and optimized by the model,but the static obstacles can't be avoided,and there is only one the point that calculated by avoidance heading and recovery timing,that doesn't meet the avoidance requirements in complex situation.