Improved Ellipsoidal Intersection Method for 2D Data Fusion

This article presents a comprehensive review and analysis of data fusion techniques for both known and unknown correlation cases. Established methods, including naive fusion, Bar-Shalom Campo, Covariance Intersection, Ellipsoidal Intersection, Inverse Covariance Intersection, Split Covariance Intersection, Expected Bar-Shalom Campo and Expected Blue Fuser are evaluated, highlighting their strengths and limitations. A key finding is that while Ellipsoidal Intersection method effectively reduces fusion covariance, it does not fully capture the common information between data. To address this limitation, this article introduces Improved Ellipsoidal Intersection method, which enhances the representation of common information by incorporating inner tangent circles alongside the ellipsoid used in Ellipsoidal Intersection method. Monte Carlo simulations under various correlation cases demonstrate that Improved Ellipsoidal Intersection method calculates lower root mean square error compared to existing techniques, particularly in high-correlation cases. The results indicate that more accurate modeling of common information significantly improves fusion accuracy. Furthermore, the implications of these findings for state estimation and multisensor data fusion are discussed, emphasizing the importance of correlation modeling in improving fusion performance. The proposed Improved Ellipsoidal Intersection method offers a promising approach for applications requiring precise and reliable data integration, particularly in cases with strong correlations.