Research On Sensor Control Strategy Based On Multi-target Multi-bernoulli Filter

With the continuous development of information fusion theory,multi-target tracking(MTT)technology has been successfully applied to many fields such as military and civilian,which makes it increasingly a hot topic in academic circles.In essence,it is a process of joint decision and estimation of time-varying dynamic systems.In multi-target tracking,it can largely optimize the level of multi-target overall tracking estimation in a complex real-world environment by developing reasonable control strategies based on the specific task planning and controlling sensor to maximize multi-target information.Mahler solves the multi-target tracking problem from the perspective of set-value estimation under the framework of finite set statistics(FISST).This approach has received much attention because it avoids significant data association processes.A series of random finite set(RFS)filters derived from this method also bring great convenience to solve sensor control strategy in multi-target tracking.In view of this,based on the FISST theory,the random finite set(RFS)filter is used to study the sensor control strategy in multi-target tracking.The main work is summarized as follows:1)Aiming at the multi-sensor control problem for MTT,this thesis proposes a multi-sensor control strategy information-based via multi-target random finite set filter.First,a multi-sensor multi-Bernoulli filter is presented by random finite set modeling,and a multi-sensor multi-Bernoulli density is approximated by a set of parameterized multi-Bernoulli processes.Secondly,by sampling the existence probability and probability distribution of the multi-Bernoulli processes separately,the sampled particle set is used to approximate the multi-target density.Subsequently,the Bhattacharyya distance,as the reward function,is used in the independent decision-making stage of multiple sensors.In addition,based on the distributed multi-sensor control strategy,this thesis also proposes a multi-sensor control strategy on the basis of multi-target tactical significance assessment,where the goal is to evaluate multi-target tactical significance and then track preferentially the maximum threat target.Finally,construct the simulation scenario and verify the effectiveness of the algorithm.2)In consideration of the sensor control problem in MTT,on the basis of the box-particle multi-Bernoulli filter,this thesis proposes an approximate solution for the sensor control strategy.First,we reasonably construct a box particle as a Gaussian distribution,and then use these Gaussian components with weights instead of the box particles obeying the uniform distribution to approximate the multi-Bernoulli density.Then,based on the two Gaussian mixed multi-Bernoulli densities,the Cauchy-Schwarz(CS)divergence is chosen as the evaluation function,and deduced in detail.Subsequently,the corresponding sensor control strategy is proposed.Furthermore,this thesis also gives a recursive formula for solving CS divergence by using point particles to approximate box particles and presents the corresponding control strategy.Finally,construct the simulation scenario and verify the effectiveness of the algorithm.