Research On Localization And Circumnavigation Problem In Multi-agent Systems With Bearing-only Information

Multi-agent system is a network system formed by a group of agents with the ability of sensing,communication,computing and execution.Encircling control is a kind of control method that the agent moves around the target or the target area according to the sensor data,It can be applied to the scene of target search,rescue,detection and monitoring.In this paper,we mainly study the multi-agent bounding control problem based on incomplete information.Among them,incomplete information means that the airborne sensors of each agent can not obtain the complete location information of the target,The incomplete position information of the target can only be obtained by a single direction or distance measurement.This paper introduces a local coordinate system calibration algorithm based on direction information and three bounding control strategies.1.Local frame calibration method based on bearing information In the distributed multi-agent system,there is no global coordinate system.The agent can only express the physical quantities sensed by its airborne sensors in their respective local coordinate systems,and realize information interaction through communication with neighboring agents.In this paper,a new coordinate system cal-ibration method is proposed,which can solve the calibration problem in the multi- =agent system only equipped with angle measurement sensor.2.Design of the algorithm of the surround control based on direction informationIn this paper,three new control strategies based on direction information are pro-posed.(1)Control strategy based on scalar estimatorThe traditional target position estimator does not make full use of the angle information.In this paper,a scalar target estimation method is designed to estimate the target distance directly.(2)Safe convex-circumnavigation control strategyConvex encirclement is the way of encircling the convex hull composed of multiple targets.In this paper,a control strategy is designed to realize that a single agent surrounds the convex hull of the target group with a specified radius,A new kind of encircling method is realized.(3)Periodically updated estimatorThe traditional target position estimator can not guarantee that the estimation of the agent and the target always keep at the same side of the target,so there is a risk of collision between the agent and the target.In this paper,we improve it and use the estimation method of periodically updating the estimation process to realize the safe encirclement which can avoid collision in the whole process.This paper focuses on the design of scalar estimator.The coordinate system calibration method can be used as a preparatory step for the following three kinds of surround control strategies.In this paper,we first propose the estimation method of scalar estimator,and propose the encirclement scheme for the relatively simple single target case to achieve secure encirclement.Then,the application of scalar estimator in complex multi-objective situation is proposed.Combined with the corresponding control protocol,a control framework to achieve multi-objective security encirclement is proposed.Finally,aiming at the deficiency of the scalar estimator in the aspect of immunity,an estimation method based on periodic update is designed,which can ensure the security and strong immunity at the same time.