Data Association for Tracking Extended Targets
Florian Meyer, Moe Z. Win · 2019
The sum-product algorithm for data association (SPADA) provides a highly effective, efficient, and scalable solution to the probabilistic data association problem, a major challenge in multitarget tracking. Similar to most existing data association algorithms, the SPADA is also based on the conventional data association assumption: Targets either produce no measurements or produce a single measurement at each time step and sensor. This paper presents the results of experiments with an extension of SPADA that is suitable for the case where targets can generate multiple measurements. This extension is general in the sense that the number of measurements generated by a target is modeled by an arbitrary truncated probability mass function (PMF) and enables extended target tracking (ETT) by performing probabilistic multiple-measurement to target associations. ETT is especially suitable for inexpensive high-resolution millimeter-wave radar sensors. We demonstrate the favorable performance-complexity tradeoff of the proposed method in a challenging tracking problem involving three closely-spaced targets that produce multiple measurements.