Research On Path Planning And Real-time Obstacle Avoidance For USVs

Unmanned Surface Vehicles(USV)are one of the autonomous products in recent years and are being widely studied and applied in practice.The planning system is one of the important systems of the USV.This article focus on planning system of USV,and uses a three-layer architecture planning system: global trajectory planning,local trajectory correction,and local real-time obstacle avoidance.Firstly,this paper introduces the research background,research significance and main content of the subject,and introduces the current research status of USV and planning system of USV,and summarizes the existing problems of existing planning algorithms.Secondly,the USV experimental platform used in this paper is introduced and part of the unmanned boat’s motion model is analyzed in depth,which lays the foundation for the design of subsequent planning system.This paper then introduces the three-layer architecture of the planning system.In the global trajectory planning layer,the digital map information is processed using rasterization.A multi-constrained multi-object genetic algorithm(IMMGA)is proposed as the global trajectory planning algorithm.The innovative points include using nondominated sorting and cost-function ranking method to deal with constraints and optimization objects,population initialization based on prior map subregions,chromosome encoding when expanding,heuristic evolution,etc.Experiments show that the IMGGA algorithm can meet multiple constraints and multiple optimization objects and planning requirements for multiple navigation scenarios.The algorithm also has advantages in terms of global track planning success rate and planning speed.In the local trajectory correction layer,a improved sparse A * algorithm(ISA-Star)based on three-dimensional space search is proposed as a local trajectory correction algorithm.The innovative points include the improvement of search space and the attributes of search node,sparse node expansion methods,etc.Experiments show that the ISA-Star algorithm can meet the obstacle avoidance requirements in various scenarios,and the results of the algorithm modification ensure compliance with unmanned sea traffic rules.The algorithm also has advantages in trajectory correction completeness and planning speed.In the local real-time obstacle avoidance layer,a 2D raster map-based lidar processing method is used,and Improved Follow the Carrot and Dynamic Window Approach(IFCDWA)is proposed.The innovations include dynamic forward-looking distance selection and dynamic window creation considering roll angle,design of evaluation function considering specific needs,etc.Experiments show that IFCDWA algorithm can effectively avoid obstacles in close range in real time,and the algorithm has advantages in track tracking accuracy and unmanned boat roll safety.Finally,this paper also uses comprehensive experiments to prove that the three-layer planning system used in this subject can well complete the planning task and meet the planning needs of USV.