Comparison between Kalman Filter and Interacting Multiple Model using 2-D trajectories

Track estimation or track smoothing is performed to minimize the errors in a tracking scenario. These errors are referred as noise in the field of tracking. Kalman filter and Interacting multiple model (IMM) filter is widely used in tracking problems since 1959. The importance and equations of Kalman Filtering (KF) are not discussed with IMM in literature although understanding KF is very important before working on IMM. In this paper the working of Kalman and its implementation in IMM is discussed along with equations in terms of time intervals t, t-1, t+1. Simulation tests were carried out and performance of IMM and KF is compared on the basis of estimation error plots. Tests are carried out on two different trajectories. KF is applied with Constant acceleration and constant velocity separately. Results indicate that KF models are not good in tracking and gives larger error rate. KF is cost effective when it comes to computations but when it comes to sensitivity of tracking, IMM is preferred to avoid error and incorrect estimates.