Multi-agent Flocking Based On Incomplete Information

The complex flocking behavior emerging in the nature generally relies on simple communication between individuals,and realizes the complex flocking behavior without global information.In fact,the study of flocking is a bionic method inspired by nature.By exploring the internal mechanism behind this complex movement,it can be applied to engineering practice and artificial intelligence to better serve human beings.Therefore,it has research value and practical significance.Based on the existing achievements,this thesis considers the impact of incomplete information on multi-agent flocking,and proposes a multi-agent flocking method based on propagandists with partially informed agents.The main achievements are as follows.(1)By analyzing the derivative of classical potential function(force function),The relationship between repulsion and attraction among agents is directly shown.This thesis proposes an improved flocking algorithm based on Morse potential function.Finally,The stability of the algorithm is proved by using the Lyapunov stability theorem and LaSalle’s invariance principle.(2)According to the improved flocking algorithm,this thesis considers the flocking problem of multi-agent with partially informed individuals based on incomplete information.Incomplete information includes three aspects.Firstly,when an agent receives information from the virtual leader or others,it is resisted or distorted information by the agents.Secondly,there is passive loss of information sent by the virtual leader or others to the agents.Finally,there is a delay in the transmission of information.Furthermore,The influence of agents with different qualities on flocking of multi-agent with partially informed individuals is discussed in this thesis.(3)Based on the improved flocking algorithm,a multi-agent flocking method based on propagandists with partially informed individuals is proposed.The propagandist is generated in the vicinity of virtual leaders.Before flocking,propagandists move regularly within an arbitrarily distributed group,disseminating information to other uninformed agents.During the whole flocking process,the propagandist is the only informed agent.This method not only solves the flocking problem of initial distribution of multi-agent with low connectivity and low density,but also reduces the unfavorable effects caused by incomplete information.Finally,in order to make the method more universal,this thesis has improved and upgraded the method again.The research in this thesis shows that.(1)the improved flocking algorithm can produce better flocking effect.(2)The flocking method based on propagandists can not only produce better flocking effect,but also effectively solve the adverse effects caused by incomplete information.Only one informed agent(propagandist)can enable most agents to track the group target.(3)The improved flocking method based on propagandists is more versatile.