Heterogeneous Fusion of an IMM Track with Measurements from Different Sources
Rong Yang, Yaakov Bar‐Shalom, Gee Wah Ng · 2019
The problem addressed in this paper is to track a maneuvering target from multiple sources, namely the source$\mathbf{S}_{\mathrm{A}}$and source$\mathrm{S}_{\mathrm{B}}$. Source$\mathbf{S}_{\mathrm{A}}$sends a target track generated by its local Interacting Multiple Model (IMM) estimator to the Fusion Center (FC), whereas$\mathrm{S}_{\mathrm{B}}$provides its measurements to the FC. The objective of the FC is to perform heterogeneous fusion of the$\mathbf{S}_{\mathrm{A}}$IMM track and$\mathrm{S}_{\mathrm{B}}$measurements. This problem cannot be solved by the existing information decorrelation algorithm [6], which works only when a (single model) Kalman filter is used in the$\mathrm{S}_{\mathrm{A}}$local tracker (suitable for tracking a non-maneuvering target). To cope with the commonly used IMM estimator (assumed deployed in$\mathrm{S}_{\mathrm{A}}$), this paper will introduce the cumulated measurement information concept, and develop the IMM Cumulated information Fusion (IMM-CFusion) algorithm. It evaluates the cumulated$\mathbf{S}_{\mathrm{B}}$measurement in the information space with multiple models, and performs fusion with the$\mathbf{S}_{\mathrm{A}}$IMM track state (without knowing$\mathbf{S}_{\mathrm{A}}$IMM “inside information” on model matched state estimates). Simulation tests are conducted to demonstrate the performance of the IMM-CFusion.