Covariance intersection in track-to-track fusion without memory

Track-to-track fusion (T2TF) can be performed under various information configurations. Under known correlations, information feedback improves the accuracy of the optimal fusion with memory, but it is detrimental to the optimal fusion without memory. This paper investigates the T2TF without memory (T2TFwoM) under unknown correlations, where the optimality is specified by the best guaranteed quality. The steady solutions are compared for three levels of information feedback. It is shown that under unknown correlations, the fusion without memory with information feedback is better than the fusion without memory with no information feedback. However, it is also shown that the fusion without memory with information feedback can be outperformed by the fusion without memory with partial information feedback in some cases. Also, the long-term optimality is analysed and some deficiencies are pointed out.