Converted State Gaussian Mixture Probability Hypothesis Density Filter for Nonlinear Multi-Target Tracking

The Gaussian mixture probability hypothesis density (GM-PHD) filter is a popular approach in solving multiple-target tracking (MTT) due to its excellent target tracking performance, such as avoiding measurement-to-track association, and its easy implementation. However, GM-PHD is exclusively applicable to linear Gaussian system models, making it challenging to be effectively utilized in scenarios involving MMT for nonlinear systems with bearing and range measurements, such as those based on radar or sonar. To solve this problem, a Converted State Gaussian Mixture Probability Hypothesis Density (CS-GMPHD) filter is proposed in this paper. Specifically, an MTT algorithm for nonlinear systems, called CS-GMPHD, is devised by combining the GM-PHD filter with an established linear Gaussian model, which is constructed by transforming the Constant Velocity (CV) model of nonlinear system from Cartesian coordinates to Polar coordinates. Finally, several simulation scenarios show that the CS-GMPHD has certain advantages in tracking accuracy and execution time, compared to the Extended Kalman filter-based GM-PHD (EK-GMPHD) algorithm and the Unscented Kalman filter-based GM-PHD (UK-GMPHD) algorithm.