Research On Swarm Formation Of Unmanned Surface Vehicles Based On Flocking Algorithm

Unmanned surface vehicles swarm formation is a system composed of multiple USVs.The unmanned surface vehicles in the system exchange information and coordinate with each other to complete specified tasks.They have the advantages of long-term operation,low cost,high efficiency and good redundancy,which can complete tasks such as maritime patrol,search and rescue,surveillance,etc.,have extremely high use value in civilian and military.For this reason,this paper studies the problem of swarm formation of USVs based on the flocking algorithm.The specific contents are as follows:1.Considering the difficulty of applying the flocking algorithm based on the particle model directly to the underactuated USV model,a conversion strategy of the flocking algorithm based on target tracking is proposed,and the target tracking controller is designed based on the backstepping sliding mode method,the flocking algorithm is effectively applied indirectly to the swarm formation of USVs.First,an agent corresponding to each USV is generated.The agent begins to converge to the flocking formation under the effect of flocking control input.Through the status information of the USV and the agent we can acquire the desired USV’s longitudinal speed and heading angle;then,for the existence of environmental disturbances such as wind and waves at the sea,a disturbance observer was introduced to compensate for the wind and wave interference;finally,the longitudinal speed and heading controller of the USV were designed in combination with the backstepping step sliding mode method,which can realize the tracking of the corresponding agent by each USV.The closed-loop system is finally consistent and bounded by the Lyapunov stability theorem.It is theoretically proved that the USV can track the agent,and it is verified the effectiveness of the algorithm by simulation experiments.2.Develop a multi-agent flocking algorithm based on variable speed virtual leader.Firstly,in order to solve the possible flocking problem of the Reynolds flocking model,the Olfati-Saber flocking model is introduced as the research basis of this paper;secondly,considering that when the virtual leader moves at a variable speed in the Olfati-Saber flocking algorithm,the flocking group converges speed is less than the speed of the virtual leader.For this,a new form of agent control input is designed to enable the flocking group to converge to the speed of the virtual leader;finally,the effectiveness of the multi-agent flocking algorithm with the virtual leader at constant speed and variable speed is verified by simulation,and combined with the target tracking conversion strategy,flocking algorithm is successfully applied to the swarm formation of USVs.3.Considering the multi-flocking behavior in nature and the existence of multi-target tracking in the actual marine operations of USVs,a method of swarm formation based on multi-flocking algorithm is proposed.First,the agent group is layered into a virtual leader layer-information layer-ordinary layer;then,in order to reduce the communication cost of exchanging state information between USVs,the information layer agent is selected using the pinning thought;finally,in order to speed up the convergence speed of the multi-flocking algorithm,the ordinary layer agent adopts the local adaptive thought to select the tracking target,and the effectiveness and versatility of the algorithm are verified by simulation.