In target tracking problems it is very important whether the abrupt change of acceleration can be detected by the the tracking filter. To detect such a maneuver several methods has been proposed. Input estimation(IE) algorithm, variable dimension filter(VDF) algorithm, interacting multiple model(IMM) algorithm are as such. Input estimation is the method that estimates accelerations considered as an input. Variable dimension filter is the method that the order of constant velocity model was augmented when the maneuver of target was detected. While this two method require a maneuver detection logic, IMM algorithm needs not such. IMM algorithm is based on the generalized pseudo Bayesian(GPB) algorithm and it is efficient in the computational point of view. But when the abrupt change of acceleration occured, it has a peak RMS estimation error because it has no maneuver detection logic.
In this paper we propose an modified IMM algorithm with maneuver detection logic to improve peak RMS error which occurs at maneuver onset time. The simulation results are given to verify that the performance of proposed method is better than that of the previous IMM algoritm.
