Research On Cooperative Target Tracking Method Based On Variational Bayes

With the rapid development of communication technology,collaborative intelligent transportation systems have become the development direction of current and future transportation systems.C-ITS can be seen as an extension of the existing intelligent transportation systems(ITS).In C-ITS,the intelligent transportation system based on vehicle-to-vehicle communication,vehicle-to-road communication and roadto-road communication can exchange information(such as the status of facilities,traffic accidents,and traffic situation)between the systems to promote the improvement of the transportation system.Effective operation to avoid accidents.Cooperative target tracking is one of the important technologies of C-ITS.By fusing the local sensing data of on-board sensors and the sensing data transmitted by vehicle-to-vehicle or vehicleto-road communication,a more accurate estimation of the target state is achieved.The effectiveness of cooperative tracking depends on the accuracy of the relative positioning between the host vehicle and the cooperative vehicle.However,the positioning signals usually provided by satellite-based navigation systems are quite susceptible to the dynamic driving environment,which affects the effectiveness of cooperative tracking.In order to achieve reliable cooperative tracking,especially when the statistical characteristics of relative positioning noise in vehicle-vehicle collaboration change over time and are uncertain,this paper focuses on the target collaboration and method in this case.(1)Research on the basic algorithm of target tracking under the variational Bayesian theory.The basic filtering algorithm of target tracking technology is introduced,with emphasis on the basic theory of variational Bayes and its application in target tracking.In-depth research on single-vehicle target tracking and collaborative target tracking has been carried out.Simulations verify that collaborative target tracking has higher accuracy for estimating the state of the target.(2)In-depth research on cooperative target tracking under the known positioning error,a cooperative target tracking framework based on Bayesian filtering is proposed.First,the state estimation formula and nonlinear observation equation linear The transformation and update formulas are deduced in detail,and then the algorithm is extended to multi-target tracking scenarios,and the pseudo-code flow of the algorithm is given.Finally,a simulation experiment based on the motion model and the Prescan platform is carried out.The results show that cooperative target tracking has better tracking effect than single target tracking,and positioning error is an important factor affecting the performance of cooperative target tracking.(3)In order to solve the limitation of the cooperative target tracking algorithm when the positioning error is known,a cooperative target tracking algorithm based on variational Bayesian inference is proposed.Because traditional collaborative tracking algorithms often do not consider the uncertainty of positioning error,but in real scenarios,the error is time-varying and uncertain,and the accuracy of positioning directly affects the effect of collaborative tracking.To solve the above problems,taking the scenario of cooperative single target tracking as an example,a variational Bayesian coordinated target tracking algorithm(VBI-CT)considering the uncertainty of positioning error is proposed.The algorithm is based on an iterative variational Bayesian framework and joint estimation The state of the target and the coordinated vehicle and the positioning noise parameters,the online variational Bayesian inference algorithm was further developed to achieve effective iterative estimation and extended to coordinate multi-target tracking.Simulation experiments are carried out based on the motion model and the Prescan platform.The simulation results show that when the positioning noise changes dynamically with time,the algorithm shows better adaptability and can effectively improve the accuracy of target tracking.