Terrain based Tracking Using Position Sensors

Tracking of aircraft and missiles has been studied in many research projects. Tracking ground targets is more difficult. Variable terrain structures limit the target kinematics and degrade the quality of measurement data. This paper focuses on tracking a single ground target that has no spatial extent. We considered the problem of ground target tracking using global and local information fused with dynamically-collected sensor data. In this context global information consists of knowledge about the sensed area, while local information deals with the target behavior and identity. The motion of ground targets is strongly influenced by general terrain and military features and other factors (eg. object mobility). In order to obtain good tracking results the topographical properties of the area have to be fused with measured data. In this work we modeled the vehicle motion using a statistical model whose diffusion in time is influenced by topographical factors modeled with a potential-field construct. We augmented this terrain based tracking algorithm (TBT), with a module that extracts direction of movement information from successive position measurements. We compared the tracking performance of the TBT with Kalman filter based algorithms for different observation rates.