Feature-Aided SMC-PHD Filter for Nonlinear Multi-target Tracking in Cluttered Environments

The Sequential Monte Carlo Probability Hypothesis Density (SMC-PHD) filter is a permissive multi-target tracker, performing state estimation through particle filtering with implicit data association. This filter is thus effective even in presence of clutter and nonlinear dynamics, while remaining tractable for real-time applications due to its computationally efficient data association process. Sensors are sometimes capable of sensing target features, which add up to kinematic measurements, e.g. range and bearing. In this paper, the adaptive Feature-Aided-SMC-PHD filter is designed, making use of feature information to increase the SMC-PHD’s estimation performance with respect to clutter, detection probability and location precision. As suspected, further differentiating targets from clutter led to greater sample degeneracy, especially as the detection probability drops. An adaptive sampling scheme was hence developed in order to relax this phenomenon. A radar application is considered in this study to validate this paper’s approach using Monte Carlo simulations.