Distributed Fusion of PHD Filters Via Exponential Mixture Densities

Abstract

In this paper, we consider the problem of Distributed Multi-sensor Multi-target Tracking (DMMT) for networked fusion systems. Many existing approaches for DMMT use multiple hypothesis tracking and track-to-track fusion. However, there are two difficulties with these approaches. First, the computational costs of these algorithms can scale factorially with the number of hypotheses. Second, consistent optimal fusion, which does not double count information, can only be guaranteed for highly constrained network architectures which largely undermine the benefits of distributed fusion. In this paper, we develop a consistent approach for DMMT by combining a generalised version of Covariance Intersection, based on Exponential Mixture Densities (EMDs), with Random Finite Sets (RFS). We first derive explicit formulae for the use of EMDs with RFSs. From this, we develop expressions for the probability hypothesis density filters. This approach supports DMMT in arbitrary network topologies through local communications and computations. We implement this approach using Sequential Monte Carlo techniques and demonstrate its performance in simulations.

Cooperative multiple dynamic object tracking on moving vehicles based on Sequential Monte Carlo Probability Hypothesis Density filter

“…The distributed fusion of information from multiple instances of the SMC-PHD filter was explored in [10] by using Exponential Mixture Densities (EMD). However, it assumes that all agents share a common FOV in which targets are detected.…”

“…We start off with a brief overview of the SMC-PHD filter and the distributed fusion of multiple SMC-PHD filter instances as proposed in [10] and proceed to introduce our extension to the current approach. A short understanding of the UKF is also provided as well as an explanation as to how uncertainties from the UKF are factored into tracking.…”

“…Our novel fusion approach generalizes the EMD method introduced in [10] by dealing with information acquired by a cooperating vehicle outside of the tracking vehicle's FOV.…”

“…In this case, our contribution involves providing a generalized approach to fuse PHD posterior intensities where FOVs of the cooperating agents are not completely overlapping. This is achieved by using the EMD approach introduced in [10] for completely overlapping segments of the field of view and a novel method for incorporating information from the cooperating agent that is outside of the tracking agent's FOV.…”

“…Bearing in mind that the fusion proposed in [10] require that two fields of view intersect, we are still left with removing particles that might be in F OV internal := F OV own \F OV other . This is achieved by determining the area defined by the intersection of two fields of view F OV own and F OV other by representing them as convex polygons.…”