Target tracking: introduction and Kalman tracking filters

The target tracking function almost invariably appears as an element of a wider system. The end-product of the system is usually a decision or a control demand, and the system may involve a high degree of human interaction or it may be autonomous. The top-level structure of such systems is usually of the form: sensor(s) → signal processing → data processing → display, decision, control. The target tracking process is usually viewed as part of the data processing sub-system and may be implemented on a general purpose processor. Signal processing is typically implemented on high speed special purpose processors. However, this separation between signal processing and tracking is somewhat artificial, and, at least for some applications, there are advantages in integrating the processes. The role of the tracking filters is system dependent, but may include several different requirements. Target tracking is a subset of general recursive estimation. An analytical solution to the general dynamic estimation problem exists for the important special case of linear models and Gaussian distributions. This is the well known Kalman filter. The paper discusses discrete and continuous Kalman filters and their role in target tracking.