Research And Application Of Multi-target Tracking In Range Test Based On RFS Theory

In the field of range measurement and control,there is a demand for the efficiency of multi-target tracking.Combined with the random finite set theory and the point set registration method,an efficient and robust multi-target tracking algorithm is proposed.Focus on the estimation algorithm for birth and death of target and the association algorithm between target and measurement and carry out systematic research work,including the following three aspects:(1)For scene of shooting range test,an efficient multi-target tracking system based on RFS theory is proposed.GLMB is not applicable in scenarios with high real-time requirements.The proposed system cancels the multi-hypothesis approach,models every target as a Bernoulli model,applies point set registration theory to the association between target and measurement,and evaluates the survival probability of each target online.So the system is a complete approach to multi-target tracking.(2)For the problem of insufficient theoretical support in the estimation of the birth and death of targets by GLMB,a life cycle model for estimating birth and death of targets based on Bayesian theory is proposed.First,a random finite set of potential targets and a surviving random finite set of targets are established.Then,survival probability is modeled as a Bernoulli distribution,and the Bayesian theory is used to estimate survival probability of each target online,so that the algorithm itself has a theoretical support.Survival state of target is determined according to survival probability.At the same time,after an association error,target can be quickly tracked again.(3)In response to the demand for a more robust and efficient association between target and measurement,an association algorithm based on point set registration is proposed.The theory of point set registration is introduced to model the association as a point set registration problem.At the same time,in order to improve the efficiency of the algorithm and be more in line with the actual needs of multi-target association.This algorithm also modifies the objective function of the original registration algorithm based on the prior information of the target motion time series,making the association more efficient and robust.For the proposed method,experimental simulation is carried out,and the results show that the method meets the high-efficiency requirements of range measurement and control.