Research And Simulation Of Pedestrian Tracking Algorithm Based On LMB Filter And Social Force Model

With the continuous development of multi-target tracking technology,pedestrian target tracking is required in the fields of traffic control,pedestrian safety,and peripheral supervision.Therefore,pedestrian modeling technology has been proposed.Previous solutions for pedestrian modeling used an independent kinematics model,which considered pedestrians as a mass point and did not integrate the interdependencies of multiple pedestrians.In order to solve this problem,this paper studies the Social Force(SF)model and realizes the modeling of pedestrians in realistic complex situations.In this model,pedestrians are not only independent particles,but also as independent entities with psychological attributes,environmental attributes,speed attributes,etc.Pedestrian movements are interdependent with other objects in the environment.Subject to personal willpower,other people's behavior,surrounding obstacles and their intended destination.In the multi-target tracking process,as the complexity of the tracking environment increases,the performance of the tracking filter algorithm is also higher,especially in the management of the target motion trajectory.In response to this problem,a delta-Generalized Labeled Multi-Bernoulli Filter(?-GLMB)was proposed,which introduces a Labeled Random Finite Set(LRFS),introduced a unique identifier for tracking targets.However,?-GLMB filtering produces multi-component quantities in iterations,resulting in increased computational complexity.In order to reduce the amount of calculation and improve the filtering efficiency,this paper uses a Labeled Multi-Bernoulli Filter(LMB),which uses an approximation method such as trajectory grouping and merging components to reduce the amount of computation.The numerical value of the iteration is small.In this paper,the social force model is integrated into a labeled multi-Bernoulli filter to form a tracking filter.The tracking filter is algorithmically studied and simulated.In order to ensure the applicability of the tracking environment,two implementation methods are introduced in detail in the research of pedestrian tracking algorithm based on labeled multi-Bernoulli filter and social force model,which is Gaussian Mixture(GM)in linear environment.Implementation method and Sequential Monte Carlo(SMC)implementation in a nonlinear environment.The simulation results show that the tracking accuracy can be obtained in the pedestrian tracking environment.