Set membership estimation applied to the localization of small UAS in tight flight formations

This paper proposes a set membership approach for UAS (unmanned aerial system)localization in a tight flight formation. A novel set membership estimation strategy based on the typical hardware available for localization (due to the cost constraints put on small UAS for civil applications) is developed. The main idea is as follows: using time-differenced differential GNSS carrier phase observations, the relative position between UAS can be tracked with centimeter-level precision, but affected by an unknown constant meter-level bias due to the initial coarse position standalone position estimate. Using pseudo range observations, as well as UWB range observations, the guaranteed space containing this position bias is determined using dense box particle sampling and sequential purging. The carrier phase trajectory fully captures the dynamics of the UAS motion and enables precise relative position holding from t = 0 on. The proposed set membership filter scheme is fully complementary to and independent of any other algorithm employed to estimate the relative position. simulation results of the problem of cooperative relative localization between UAS on a formation flight benchmark compared to a standard Extended Kalman Filter illustrate the benefits arising from the deterministic nature of set membership filtering.