Covariance Intersection in Track-to-Track Fusion: Comparison of Fusion Configurations

Tracking and localization are typical application fields of estimation fusion algorithms. Multiple sensor nodes can collaborate by sharing estimates of, e.g., the position and velocity of a target. This paper compares several fusion configurations, which are distinguished by the communication rate, the information feedback type, and the use of memory. The attention is focused on communicationally cheap suboptimal solutions that avoid the computation of the cross-correlations of estimation errors by considering all admissible cross-correlations. The fusion configurations are compared based on the quality that can be guaranteed by the used fusion rules.