Path Planning And Collision Avoidance For Unmanned Surface Vehicle In Complex Marine Spatial Environment

With the deepening of marine research and development,traditional shore-based/space-based observation and development methods have been unable to meet current requirements about the ocean field,and unmanned marine platforms have gradually developed into an important way to make up for the de ficiencies of traditional methods.As the only surface unmanned platform on o cean,unmanned surface vehicles(USVs)have received more and more attention,and have been applied in many industries.Technologies of USVs involve multiple research areas and represents a frontier of high technology.However,the autonomous collision avoidance technology is still the main bottleneck to the automation of USVs.Meanwhile,USVs’missions are inseparable from the path planning,and the implementation of appropriate global path planning under different mission requirements can effectively improve the efficiency of USVs’missions.Therefore,the research of global path planning and local collision avoidance technology for USVs has theoretical and practical significance.In this dissertation,the global path planning and local collision avoidance for USVs are focused on,and the main contents of this dissertation are summarized as follows:For the research of global path planning,aiming at the problem of poor environmental adaptability of parameters of the conventional Fast March Square(FM~2)method,an environmentally adaptable global path planning algorithm with an inshore distance constraint fast marching square(IDC-FM~2)method is proposed.The IDC-FM~2 method uses two inshore distance parameters to constrain the path portions bypassing obstacles within the area determined by the two parameters,and obtains a quasi-time-optimal and smooth global path while ensuring the safety of USVs.Aiming at the problem of poor calculation efficiency of global path planning based on grid maps with large scale,a rapid global path planning algorithm based on two level spatial resolution grid maps is proposed.This algorithm plans paths twice based on two level spatial resolution grid maps.For the first time,a preliminary path planning is performed on a low spatial resolution grid map to determine effective regions where the effective path locates in with high probability distribution.For the second time,a finer path planning is performed on the effective area of the high spatial resolution grid map.Based on the above processing,the calculation efficiency can be effectively improved without sacrificing the accuracy of the planned path.For the local collision avoidance in USV applications with requirements of long voyages and long distances,a local collision avoidance algorithm based on independent steering is proposed.The algorithm aims at the requirement of maintaining economic speed navigation to reduce fuel consumption under applications with long voyages and long distances,and avoid collisions only by steering operation.This algorithm evaluates collision risks through a risk assessment model,introduces a concept of steering occasion which is characterized by distance parameters,and determines whether to perform collision avoidance operation based on the steering occasion.Under the common constraints of an analytical navigable angle interval and International Regulations for Preventing Collisions at Sea(COLREGs),a navigation waypoint which meets the collision avoidance requirements by steering is generated to guide the USV.On this basis,the entire local collision avoidance is managed and executed through a finite state machine model.For the completeness requirements of global path planning and local collision avoidance in the navigation of USVs,a hybrid path planning algorithm is proposed by introducing the path re-planning technology to combine the global path planning and local collision avoidance organically.The path re-planning technology is obtained through adaptive improvement under different situations based on the basis of global path planning.The algorithm performs global path planning once before the task is executed,and then a USV is navigated by tracking the planned global path.When the USV encounters dangerous obstacles and goes to perform collision avoidance operations during the navigation,a temporary waypoint is used for the individual steering collision avoidance,which is determined by organically combining the global information introduced by the path re-planning technology and local individual steering collision avoidance method,or the collision avoidance with both varying speed and steering is performed in emergency collision avoidance situations.When the USV is out of the risk of collision,the remaining quasi-optimal path is re-planned by the path re-planning technology,and the planned path is tracked.Meanwhile,aiming at the the large number of pre-simulation verification requirements for the proposed hybrid path planning algorithm,a semi-physical simulation system is designed,which is composed of two real systems of USVs(the USV main control system,and the information displaying and control with communication system)and two virtual simulation systems(USV motion simulation system and obstacle simulation system),and used to provide a software and hardware system verification platform as close as possible to the actual USV operating environment for the proposed hybrid path planning algorithm.