Researches On Consensus Problems Of Multi-agent System With Information Package Dropouts And Anchors’ Selection Problem

Multi-agent system’s consensus refers to the final states of the agents reach their consensus through mutual cooperation, which involves many problems such as consensus of multi-agent systems with uncertain communication, formation and obstacle avoidance in the multi-agent systems to obtain it’s consensus in dynamic topology, etc. And it has been widely used in the fields such as information fusion, cooperative control etc. So the multi-agent’s consensus receives much attention from many researchers both at home and abrod in various fields.Two main aspects was studed in this paper: 1, Consensus problems of multi-agent system with information package dropouts and further studies on communication trends in this multi-agent system. 2, Fault tolerant problem of anchor agents’ selection in multi-agent system and anchors’ selection problem in 3-D space while positioning.Firstly, consensus problems of multi-agent system with information package dropouts and further studies on communication trends in this multi-agent system. The agents’ equipment will drop packets inevitably in the actual multi-agent’s communication process. In order to tackle this problem, this paper proposed one method that via hold to keep on communicating, which utilized the historical communication state information to substitute the failure communication state information of the agent so as to ensure the system run smoothly in a large extent. What’s more, a new communication gradient operator was defined for studying the communication trend based on the multi-agent reaches their consensus.Secondly, fault tolerant problem of anchor agents’ selection in multi-agent system. Multi-agent in certain formation and area should determine their position before reaching their consensus. Formation formed by agents with distance sensors should be globally rigid and some anchors which know their coordinates should be chosen out when positoning, the other agents can be located via the constraint relations between anchors and other agents in the formation. Because of force majeure, anchors’ s failure will lead the whole muti-agent system to be paralyzed. In order to improve the multi-agent system’s stability, this paper presents one fault tolerant mechanism from anchor selections’ point of view when self-localization in multi-agent autonomous formations are available to the agents in a globally rigid formation, by a way that converting the single set of the optimal anchor selection problem into the multi-sets of the sub-optimal anchor selection problem, which guarantees that system for stable operation. Furthermore, the concept of influence ratio of the failure anchor was coined to cover the importance of the corresponding anchor, which facilitates the maintenance of anchors in the multi-agent system. Base on the above reaserch, to avoid strict requirements that the agents in original autonomous formations be globally rigid formation when only distance measurements are available in the localization of multi-agent systems, the paper proposed volume coverage to select anchors in the multi-agent systems, this algorithm doesn’t require the original formation to be globally rigid. This method can be utilized to select anchors while it is under less demanding conditions, meanwhile, this paper considers that selecting the anchor via this method in 3-D space.