A new maneuvering target tracking algorithm with input estimation

A new tracking algorithm is proposed. It treats the target acceleration as a nonrandom term, and consists of a constant velocity filter an input estimator and a maneuver detector implemented in parallel. The new method has the same advantages as the two-stage Kalman estimator which requires a lesser amount of computation and provides even a better performance when compared with an augmented state Kalman filter. At the same time, the new method uses a better tuning parameter and removes a difficulty in implementation of the two-stage Kalman estimator. It is shown that the new filter is a better alternative to the two-stage Kalman estimator on tracking maneuvering targets.