What flies over one house

Delivery drones broadcast their position and identity continuously, by law. These are the ones that passed over a single address in Texas, recorded by our own receivers — not a national feed, and not anyone else's hardware.

A detection is one broadcast received — not one aircraft, and not one delivery. A drone overhead transmits every second or so, so a single trip across the sky produces dozens. That is why the aircraft count matters as much: one drone flying all day can out-count fifty that pass once.

2026-09-07 to 2026-10-07.

When they fly

Hours are UTC until this page knows your timezone.

What flies

Read both number columns together. Amazon appears far more often than anyone else, but from very few aircraft — the same handful of drones, back and forth, all day. Wing is the opposite: many more aircraft, each seen only a few times. Two operators can log similar detection totals and be running completely different operations, and only the aircraft column shows it.

Across the wider network

Everything above came from our own receivers at one address. This section is wider: it adds every other receiver whose data reaches us, most of them run by other people across the region. That covers far more ground, but we know less about it — the receivers are scattered unevenly, and a drone that never flew near one does not appear here at all. Read it as "what got picked up", not as "what flew".

The same applies here. Zipline and Wing log far fewer detections than Amazon but fly many more individual aircraft — a fleet spread across a city, rather than a few drones working one route.

Which radio they use

Remote ID can be broadcast over Bluetooth (BLE) or Wi-Fi — the standard permits either, and leaves the choice to whoever built the aircraft. In this data nobody mixes them. Each type picks one radio and uses it for every broadcast.

Those two columns answer different questions. Bluetooth leads on broadcasts and loses badly on aircraft, because almost all of it is one operator's small fleet transmitting all day. Counted by airframe rather than by broadcast, Wi-Fi is the one nearly everything uses.

The obvious objection is that this is a map of our antennas, not of the aircraft. A Bluetooth receiver sitting under one operator's route and a Wi-Fi receiver somewhere else would produce this table out of nothing. So the same question was put to the 4 receivers that recorded both radios in this window — same hardware, same place, listening to Bluetooth and Wi-Fi at once. 42 aircraft types passed them, and not one of them used both.

53,606 of 81,449 detections (66%) came from receivers that record which radio they heard; the rest report only that they heard something, and are left out of this section entirely. One further limit worth stating: our own Wi-Fi receiver listens on a single channel, so if this table is wrong in a predictable direction, it is undercounting Wi-Fi.

How this was recorded

An aircraft is counted once, by the serial it broadcasts. Every aircraft in this class transmits an identifier, and a single drone is one aircraft here however many times it passes. Contacts where that identifier was not decoded — a garbled frame leaves only a hardware address — are shown in the detection totals but not counted as separate aircraft, because there is no way to tell such a fragment from a drone already counted.

"When they fly" and "What flies" come from receivers we run ourselves, at one address. "Across the wider network" comes from the regional network, most of which is other people's. It is a description of one patch of sky, not of a city or a country: an aircraft that never came within radio range is simply absent, and a quiet hour may mean nothing flew or may mean nothing was heard. No positions, identifiers or receiver locations appear on this page. The receiver is a Raspberry Pi and about $100 of parts if you want to record your own.