Tracking multiple point targets using genetic interacting multiple model based algorithm

Abstract

Multiple point target tracking in the presence of dense clutter requires tracking maneuvering and non-maneuvering targets simultaneously in the absence of any apriori information about target dynamics. We propose a tracking algorithm based on interacting multiple model (IMM) which exploits the genetic algorithm for data association. In the proposed algorithm no observation is assigned to any trajectory, but assignment weight is calculated using the genetic algorithm for validated observations for each trajectory. For inclusion of multiple models, the likelihood of an observation is modelled by mixture probability density function (pdf). The proposed algorithm provides robust data association and inclusion of different dynamic models for the target allows one to track an arbitrary trajectory.