Research On Path Planning And Track Control Technology Of Unmanned Surface Vehicle

As a highly autonomous surface ship,the unmanned surface vehicle has the characteristics of small size and high flexibility.It can perform tasks that are particularly dangerous,boring and not suitable for manned ships.It has great development prospects in the fields of civil,military and marine exploration and development.However,effective navigation and accurate track control is the key to complete specific tasks.Therefore,the global path planning method,local dynamic collision avoidance method and path control method are studied in this thesis.The main work of the thesis is as follows.(1)A six degrees of freedom(6-DOF)maneuvering model of unmanned surface vehicle(USV)is established,and then a three degrees of freedom(3-DOF)maneuvering model including nonlinear damping term is derived according to the underactuated and uncertain characteristics of USV,which provides conditions for the following research on global path planning,local dynamic autonomous collision avoidance and path tracking control.(2)In order to solve the problem of long planning time caused by repeated searching in the process of searching the optimal path of global path planning for unmanned surface vehicle,a global path planning algorithm based on A* algorithm is proposed.Firstly,the occupied grid method is used to grid the known navigation map.Secondly,A* algorithm combining double search list is designed to search the shortest path to the target.There may be other unknown environmental information in the navigation area.A Hector simultaneous location and mapping algorithm is designed to estimate the location and scan the grid map.Finally,the standard of path re-planning is introduced to solve the problem of real-time path re-planning.Simulation results show that the proposed method can quickly find the shortest path to the target while sailing,and can re-plan the path in real time under certain conditions.(3)Aiming at the problems of oscillation and target deadlock in dynamic collision avoidance of multiple unmanned surface vehicles,a dynamic autonomous collision avoidance method based on improved integrated environment algorithm is proposed.Firstly,a circular sensor disk area is designed in front of the USV,in which measured environmental information is integrated,and safe and effective collision avoidance strategies are selected according to the state information of obstacles in the disk.Then,a braking rule similar to the yield principle is designed.In the case of multiple unmanned surface vehicles meeting,the collision avoidance strategy selected according to the braking rule can make each unmanned surface vehicle reach its target safely.Simulation results show that the method can effectively avoid various obstacles in various scenes.(4)Aiming at the problem that it is difficult to establish accurate dynamic model and the control law obtained by traditional algorithm is very complex and difficult to realize engineering practice,a path control algorithm of unmanned surface vehicle based on deep reinforcement learning is proposed.Firstly,a deep reinforcement learning algorithm based on deep deterministic policy gradient is designed,which can achieve accurate trajectory control without the need for an accurate USV control model.Then a double Gauss reward function is designed,which integrates the navigation state information of the USV to evaluate the training actions,so that the proposed algorithm can track the given path faster.The simulation results show that the proposed algorithm has better control performance than the explicit model predictive control and linear quadratic regulator.