Research On Dynamic Collision Avoidance Of Unmanned Surface Vehicle Based On Particle Swarm Optimization

As the advantage of Unmanned Surface Vehicle(USV)has been concerned about gradually,the expectation of a significant improvement in the intelligrnt level of USV and its widely application in military and civilian fields is rising day by day.In the field of research involved in USV,the ability of autonomous planning is an important reflection of its degree of intelligence.Dynamic collision avoidance is a key part of it,which aims to keep the USV and the dynamic obstacles within a safe range during travel.This capability is the basic safety guarantee for USV to be able to complete the assigned tasks on the water.In order to study the problem of dynamic collision avoidance of USV,this paper constructs a collision risk index model,presents a method for judging dynamic collision avoidance situation and proposes an improved particle swarm algorithm method.Aiming at the intelligent requirements of USV,based on the constructed safe distance of approach model,calculating the collision risk of USV based on fuzzy rules and fuzzy comprehensive evaluation,designing Compound Collision Risk Index,which reflect the dangerous relationship between the USV and the obstacles from different angles.Finally,fuzzy language values are used to describe the collision processBased on the analysis and classification of the relationship between the various hazards of different dynamic obstacles,the situation judgment of the dynamic collision avoidance of the USV is given,and the basic rules and processes under the normal maritime rules avoidance,emergency avoidance and rerouting are determinedAiming at the collision avoidance strategy of the USV,a speed-heading strategy based on the collision avoidance tow point is proposed.Attraction and Escape Switched PSO based on Von Neumann topology is proposed.By constructing fitness functions in different situations,the algorithm’s ability to adapt to complex situations is guaranteed.Simulation experiments in different environments verify the timeliness of the algorithm.