Research On Multi-agent Collaborative Positioning Algorithm

In recent years,with the development of all kinds of fields,intelligent robots,unmanned aircraft,unmanned vehicle,the demands of high location precise become more common.Compared with the traditional single intelligent agent using more sensors or improving the precision of the sensor that costs more time and money,this paper puts forward two kinds of collaborative network positioning algorithm,which use multiple intelligent agents that carry combination positioning system to communicate with each other,to improve the positioning precision and anti-jamming capacity of each single agent:(1)To improve the positioning accuracy and anti-interference ability of individual intelligent agents,a network positioning algorithm based on Federated filter structure was designed.algorithm uses the combination positioning system of GPS/INS loose coupling structure as a network of filter layer.Then using the coordinates of each node in the network and its exploration of other nodes' distances to build the central filter layer.The final layer uses the central filers' results of all the agents in the network as the federal filter's inputs to calculate the final optimized filtering results.The simulation results show that Federated filter structure positioning system has a better effect on the positioning accuracy than the GPS/INS positioning system of a single intelligent agent.(2)In order to detect the individual failure and attack mode,the algorithm of fault detection based on the algorithm of clustering algorithm OPTICS and the recognition algorithm to deal with the generated deception and forward spoofing are studied.The simulation results show that Federated filter structure positioning system has a better effect on anti-interference ability than the GPS/INS positioning system of a single intelligent agent.(3)In order to deal with the difficulty to build the fully connected network positioning system,and cannot deal with the change of agents' number,The Ad-hoc network positioning system using ad-hoc network structure is proposed.System uses the connection matrix and Dijkstra algorithm to solve the problem of the shortest jump path in ad-hoc network communication.And the improved filtering algorithm for the matrix dimension change in ad-hoc self-organization is proposed.Simulation shows that the Ad-hoc network positioning algorithm can keep a high positioning accuracy and good anti-jamming with the limitation of the network communication.