Feature-Aided Tracking of Ground Vehicles using Passive Acoustic Sensor Arrays

Tracking of a moving ground target using acoustic signals obtained from a passive sensor network is a difficult problem as the signals are contaminated by wind noise and are hampered by road conditions, terrain and multipath, etc., and are not deterministic. Multiple target tracking becomes even more challenging, especially when some of the vehicles are light (e.g., wheeled) and some are heavy (e.g., heavy wheeled vehicles like trucks, tracked vehicles like tanks, etc.). In such cases the stronger acoustic signals from the heavy vehicles can mask those from the light vehicles, leading to poor detection of such targets. The full position estimates of emitters (targets), obtained following the association of the DoA angle estimates from multiple sensor arrays at each time scan, are used for target tracking. However, because of the particular challenges encountered in multiple ground vehicle scenarios, this association using kinematic (DoA angle) measurements only is not always reliable and can lead to lost as well as false tracks. In this paper we propose a new feature-augmented static association algorithm where feature augmented DoA angle measurements from multiple sensors are associated to localize targets and obtain composite measurements (position estimates) using a static multidimensional assignment (MDA) framework. We present a novel DoA detection scheme followed by a feature extraction technique designed from and for real data. Dynamic S-D and feature-aided S-D (multidimensional) assignment algorithms are presented to assign composite measurements and feature-augmented composite measurements, respectively, to tracks. The techniques are developed based on real data sets and tested on real data based on a field experiment.