Multisensor Maneuvering Target Fusion Tracking Using Interacting Multiple Model
Abstract
For multisensor maneuvering target tracking, two important factors affecting the tracking performance are: (1) the uncertainty of the target dynamics model; (2) the cross-correlation of local estimation errors across sensors. For these problems, a new model-level information fusion algorithm based on interacting multiple model (IMM) is proposed. First, in each local sensor, the IMM algorithm is used to deal with the problem of uncertainty of the dynamics model caused by the target maneuver, and the obtained model-level information (Gaussian mixture probability density) instead of the state estimation after model mixing is sent to the fusion center. This effectively avoids the loss of information in the process of model mixing. Second, for the correlation between local estimates, a new model level information decorrelation algorithm for IMM is proposed to obtain decorrelated fusion information. Finally, in the fusion center, the fusion of the de-correlated estimation information is completed by the naive fusion method. The simulation experiments verify the performance of the proposed algorithm.
