Research On Path Planning Algorithm Of USV

With the rapid development of technology,unmanned equipment level continues to improve,the USV play a prominent role in military and civilian fields,and its development has been paid more attention to.Autonomous navigation is a key technology for USV and toward modular,systematic,intelligent direction.USV needs high adaptability and intelligence in carrying out its mandate,automatically plan a safe,economic and efficient path in a complex marine environment.Path planning is divided into global path planning and local path planning,research on global path planning which is relatively mature.Based on the global path planning,research on local path planning in dynamical environment,planning a local path to avoid dynamical obstacle and meet with COLREGS.This paper first briefly introduces the research status and development trend of domestic and foreign unmanned surface vessel;then analyzes the COLREGS the USV should comply and agent control architecture.In this paper,apply the hybrid architecture framework,in deliberative layer using a grid method for modeling the static marine environment,use A Star algorithm to plan a global path and sent planned waypoint information to the reaction layer;The reaction layer,apply perception-planning model to local path planning,although the USV can avoid moving object,but is not necessarily meet the COLREGS,easily make impact on other ships safely navigation.Adding a parameter to dynamic avoid collision behavior,build the behaviors comply the rule.Using multi-objective optimization algorithm to plan a local path based on behavior;use IvP to approximate the objective function;search through the decision space by branch and bound method,improve the search efficiency,rapidly generate control decision.Finally,develop and optimize open source software,implement the USV sailing along the planned global path,design the head-on and crossing situation.Conduct simulation experiment,the result consistent with COLREGS,verify the validity of model and reasonableness of the algorithm.