Federated Learning for Unmanned Aerial Systems: A Screened Bibliometric Census of an Emergent Sub-Literature

We census the literature at the intersection of federated learning (FL) and unmanned aerial systems using a Scopus query combining FL terminology, UAV terminology, and delivery/logistics terminology (COMP and ENGI subject areas). The exported raw retrieval contains 14,965 records. Two screening steps take that number to the corpus we actually analyze: 39.6% of the raw records mention FL-related terminology (federated or distributed learning) anywhere in title, abstract, or keywords, and of those, only one in five additionally mentions UAV-related terminology, a joint match rate of just 7.9%. Screening on that joint criterion yields a final analytic corpus of 1,185 records (2022–2026), still expanding with no sign of saturation (90 papers in 2022, 361 in 2025, 332 already indexed for a partial 2026); we validate this figure with FL- and UAV-side term-sensitivity checks and a manual read of a random sample, which finds roughly 54% of screened records substantively about FL-for-UAV rather than mere keyword co-occurrence. IEEE Internet of Things Journal leads by volume (111 papers, $h=30$); author productivity follows Lotka's law almost exactly ($n=2.33$, $R^2=0.998$, 2,820 authors); China leads by volume (628 country-affiliation instances) while Germany, Australia, and Hong Kong lead by per-paper citation impact. A cross-check against the identical screening criterion applied to an independent OpenAlex export finds the same literature at roughly one-eighth the volume (147 records), a coverage gap consistent with Scopus's broader proceedings indexing. We report a further, practically consequential finding for anyone scoping this space by keyword: of the 228 records containing a delivery-adjacent term, only 5.7% mention a parcel-specific term (parcel, last-mile, courier); the rest reflect transportation, traffic, and generic-logistics engineering idiom, not parcel logistics, so keyword screening alone cannot isolate last-mile drone-delivery research from the surrounding UAV-networking literature, and the delivery-specific sub-domain remains, in the indexed record, too small and too entangled with communications terminology to census on its own using reproducible bibliometric methods.