Research On Cooperative Strategies Of Distributed Multiple AUVs For Searching And Hunting

Underwater Vehicles(Autonomous Underwater Vehicle/AUV,Unmanned Underwater Vehicle/UUV),as powerful and useful tools in ocean development,exploration and confrontation with capabilities of autonomous control,decision and navigation,have been a hot research topic in robotics area.Compared with individual underwater vehicle system,Multiple Underwater Vehicles have advantages in efficiency and robustness due to the distributed characteristics in time and space.However,cooperative strategies needed in distributed multi-vehicles systems also bring difficulties in control system design process.As a consequence,in this dissertation,regarding searching and hunting mission for multi-vehicles system,distributed cooperative control methods are investigated concerning about three sub-missions including position-guide maneuvering,area searching and target hunting.The details are listed as follows.1.System architecture decides the framework of cooperative control system.As a consequence,distributed system architecture of multi-vehicles systems is developed at the first place.With the communication layer as the physical basis of cooperation among vehicles,mathematical presentations of communication topology and general cooperative control system are constructed based on graph theory.In addition,considering about the complex dynamics of underwater vehicles,feedback linearization method is adopted to obtain second-order integral model of the system.2.Concerning about the position-guide maneuvering mission of multi-vehicles systems,with thorough analysis of demands,the mission is mathematically formulated as a cooperative formation control problem.With consideration of the discrete property brought by communication system,based on the discrete cooperative control system,the convergence of formation control system is analyzed under various conditions.Through establishing the equivalent relationship between formation control and consensus control with fixed formation,stochastic matrix theory and interval matrix theory are introduced to obtain the sufficient conditions for consensus of the system(1)with no communication delays,(2)bounded communication delays and(3)unbounded communication delays.With the simulation experiments for 2D surface case and 3D underwater case,effectiveness of the proposed methods are demonstrated.3.Concerning about the coverage searching problem in area searching missions.A task partitioning method is proposed inspired by competition rules among animals.With consideration of threat map and different abilities of individual vehicles in system,a dynamic model depicting task burden of each vehicle is constructed by imitating the competition process for resources and territories in wild nature.By application of consensus theory,it is proved that the consensus in task burden for each vehicle is guaranteed for coverage searching missions.4.Concerning about the targets detecting problem in area searching missions.With consideration about the priorities of targets,simulated annealing algorithm is used to solve the priority traveling salesman problem constructed in order to acquire the path planning method for each vehicle system.On the basis of Pareto optimality in game theory,the definition of Pareto consensus is put forward for task allocation problem.With the proposed definition and K-means clustering,a distributed negotiation algorithm is developed to settle the target detecting missions for multi-vehicles system.5.Concerning about the target hunting mission of multi-vehicles systems,the problem is divided into two sub-problems according to the strategies adopted by the target including cooperative target following and pursuit-evasion game.For the cooperative target following problem,without any information about the acceleration of the target,distributed adaptive controller is designed at first with convergence in target following and state estimating processes.With the proposed controller structure,the situation where only delayed information of target is available is studied.In this case,through separating the multi-vehicles systems into active following subsystems and passive following subsystems,and based on Lyapunov-Krasovskii functions,control strategies for different subsystems are presented with sufficient conditions needed to guarantee the boundedness of target following errors.For the pursuit-evasion game,a design method about cost functions for distributed systems is proposed which both keeps the physical meanings of the functions and ensures the Nash equilibrium characteristics of the distributed control strategies of each vehicle.Furthermore,analysis about the impacts of chosen parameters in cost functions on the pursuit-evasion game is carried out with simulation tests.