I logged every delivery drone over my house for 16 days…
0 reactions
I logged every delivery drone over my house for 16 days. Amazon flies 26 aircraft 345 times each; Wing flies 218 aircraft 9 times each.
Delivery drones broadcast their position and identity continuously — it's required
under ASTM F3411 Remote ID. You can receive it with a Raspberry Pi, a Bluetooth
adapter and a Wi-Fi adapter that does monitor mode. No interception, nothing private;
it's a public broadcast aimed at exactly this.
I've been recording it for 16 days. 14,375 detections of 544 individual aircraft.
(A detection is one broadcast received — a drone transmits every second or so, so one
trip overhead produces dozens.)
The thing I didn't expect was what happens when you count \*airframes\* instead:
**Operator**|**Detections**|**Distinct airframes**|**Per airframe**
Amazon Prime Air MK30|8,963|26|**345**
Zipline P2|2,294|119|19
Wing Hummingbird|1,948|218|**9**
Flytrex Sky II|236|30|8Amazon flies 26 aircraft regularly. Wing flies 218 aircraft occasionally each.
Those are the same business on paper and completely different operations in the air —
Amazon concentrates a tiny fleet on repeated corridors, Wing spreads a large fleet
thinly across store-based sites. You can't see that in a detection count alone, and I
haven't found it published anywhere.
Time of day was the other surprise. Traffic starts around 07:00, peaks at 10:00 and
again 13:00–17:00, and stops before 19:00. It's a working-day logistics pattern, not
an evening food-delivery one.
\*\*Limits, because they matter:\*\* this is one suburb in north Texas plus a wider
receiver network, and coverage is uneven. An aircraft that never came within radio
range is simply absent. A quiet hour might mean nothing flew, or might mean nothing
was listening. The per-airframe numbers are the most robust thing here, because
they're ratios within each operator rather than comparisons across coverage.
Happy to answer questions about the receiver side — Pi 5, a Bluetooth 5 dongle and an
Alfa AWUS036ACM. The fiddly part is monitor mode, not the decoding.
\*Disclosure: the data comes from a tracker I built. Link in a comment if anyone wants
it, rather than in the post.\*
#technology
source
Delivery drones broadcast their position and identity continuously — it's required
under ASTM F3411 Remote ID. You can receive it with a Raspberry Pi, a Bluetooth
adapter and a Wi-Fi adapter that does monitor mode. No interception, nothing private;
it's a public broadcast aimed at exactly this.
I've been recording it for 16 days. 14,375 detections of 544 individual aircraft.
(A detection is one broadcast received — a drone transmits every second or so, so one
trip overhead produces dozens.)
The thing I didn't expect was what happens when you count \*airframes\* instead:
**Operator**|**Detections**|**Distinct airframes**|**Per airframe**
Amazon Prime Air MK30|8,963|26|**345**
Zipline P2|2,294|119|19
Wing Hummingbird|1,948|218|**9**
Flytrex Sky II|236|30|8Amazon flies 26 aircraft regularly. Wing flies 218 aircraft occasionally each.
Those are the same business on paper and completely different operations in the air —
Amazon concentrates a tiny fleet on repeated corridors, Wing spreads a large fleet
thinly across store-based sites. You can't see that in a detection count alone, and I
haven't found it published anywhere.
Time of day was the other surprise. Traffic starts around 07:00, peaks at 10:00 and
again 13:00–17:00, and stops before 19:00. It's a working-day logistics pattern, not
an evening food-delivery one.
\*\*Limits, because they matter:\*\* this is one suburb in north Texas plus a wider
receiver network, and coverage is uneven. An aircraft that never came within radio
range is simply absent. A quiet hour might mean nothing flew, or might mean nothing
was listening. The per-airframe numbers are the most robust thing here, because
they're ratios within each operator rather than comparisons across coverage.
Happy to answer questions about the receiver side — Pi 5, a Bluetooth 5 dongle and an
Alfa AWUS036ACM. The fiddly part is monitor mode, not the decoding.
\*Disclosure: the data comes from a tracker I built. Link in a comment if anyone wants
it, rather than in the post.\*
#technology
source
| 0 mlx | |
| 1 | |
| 0 |