Diver tracking in unknown structured clutter background using a force-based GM-PHD filter

This paper considers the problem of tracking multiple human divers using a high-resolution 2D active sonar. The proposed solution can handle the real-world challenges of a time-varying number of targets, complex correlated diver dynamics affected by external factors (e.g., water current, activities of neighbouring divers and the intent of the divers themselves), and unknown spatial non-homogeneous clutter intensity due to the complex non-stationary underwater environment with structured clutter. The proposed algorithm uses the probability hypothesis density (PHD) filter with a novel force-based diver state evolution model, along with a log-Gaussian Cox process (LGCP) model for a global clutter spatial intensity estimation. In addition, the posterior Cramér-Rao lower bound (PCRLB), which quantifies the best possible accuracy in the presence of kinematic interactions among targets in an uncertain underwater structured environment, is derived as the benchmark for performance evaluation. Simulation results demonstrate the improved performance of the proposed method over the standard PHD filter.