Design of the adaptive interacting multiple model algorithm

Adaptive interacting multiple model (AIMM) estimation is a method to improve the interacting multiple model (IMM) estimation. But some new problems appear in the AIMM, such as how to select the structure of the adaptive model set, how to inherit the different datum of the filters based on the old model set. In the paper some instructive conclusions and formulas of the design of the model set M and the model transition probability are presented to solve these problems. A new turn rate estimation method based on a pseudo-observation is given. Several new AIMM algorithms are put forward based on the estimation. Simulation results show the better performance of these new AIMM algorithms.