Multi-target tracking for SMARTnet: multi-layer probability hypothesis filter for near-Earth object tracking.

Frueh, C.; Fiedler, H.; Schildknecht, T.; Herzog, J. (2021). Multi-target tracking for SMARTnet: multi-layer probability hypothesis filter for near-Earth object tracking. In: 8th European Conference on Space Debris. Darmstadt, Deutschland. 20. April 2021 - 23. April 2021.

Text TS_ECSD2021.pdf - Published Version Restricted to registered users only Available under License Publisher holds Copyright. Download (455kB)

n this paper, a modified version of the finiteset statistics-based Probability Hypothesis Density(PHD) filter is developed specifically for the opticalmulti-target tracking of objects in the near-Earthrealm for Space Situational Awareness (SAA). Atwo-step PHD filter is proposed in a modified ver-sion. One labeled PHD filter is used on the orthogo-nal image plane, in which linear dynamics in a four-parameter state is employed, forming so-called track-lets. Tracklets are associated sets of a few closely-spaced observations covering a negligible part of theoverall orbit. Furthermore, tracklets are fed into asecond PHD filter in a modified measurement updateversion, utilizing the full near-Earth astrodynamicswith a six parameter state. In the modification, eachtracklet leads to only one update in the PHD, but allobservations within the tracklet are processed in thesingle target Markov transition process within thefilter. In this case, the single target filter is an Ex-tended Kalman Filter. In addition, the birth processthat has been usually in typical SSA applicationsshifted to the birth step, forcing a data-driven birthwith the disadvantage of a severe model mismatch,back to the propagation step, as in the original PHDfilter formulation, avoiding the mismatch. In order toovercome the lack of probabilistic description avail-ability (one of the triggers of the shift to the data-driven update step of previous authors), the data ispreprocessed. This has the advantage that birth canemploy traditional initial orbit determination meth-ods and does not have to rely on the initializationwith an incomplete state using, e.g., an admissibleregions approach. The results are generated usingthe optical data of the DLR SMARTnet telescopenetwork and are compared to the DLR BACARDIdata processing.

Interest & Impact

Downloads

2 since deposited on 18 Feb 2022
0 in the past 12 months

Citations

5 Citations in Web of Science ®
6 Citations in Scopus

Search

in Google Scholar™

Services

Actions (login required)

Edit item

Actions (login required)

Edit item