Department of Biomedical Engineering and Computational Science

Tracking a Target with Simple Manouvers, 'imm_demo'

In this demonstration program the IMM filter is used in estimating the trajectory of a target, whose dynamics are modeled with the following two alternating models: the standard Wiener process velocity and acceleration models. The purpose of the latter is produce sharp turn to object's trajectory, while the former keeps the trajectory stabile.

Figure 1 shows the filtering and smoothing results of the IMM filter and smoother using the two models mentioned above. See the documentation for discussion and performance comparison to classical Kalman filter and smoother.

The IMM filter provides estimates for model probabilities on each sampling period. Figure 2 shows the filtered and smoothed probability of Wiener velocity model on each time step. Notice how the lag in the filtered estimates are recuced by the application of smoother.

Files used in this example:

IMM_DEMO Tracking a Target with Simple Manouvers demonstration
Figure 1

Figure 1. Filtered and smoothed estimates produced by the IMM filter and smoother.

Figure 1

Figure 2. Estimated probability of the Wiener velocity model on each sampling period.