Detect drones that don’t want to be seen.
A mesh of passive sensors, built from commercially available parts, that hears drones, fuses the contacts, and triangulates them.
Today’s systems only see the aircraft that announce themselves.
Remote ID and ADS-B assume a compliant operator. The threat broadcasts nothing. That is the gap.
Listen. Fuse. Locate.
Signatures it cannot hide
Each node passively picks up rotor sound and the RF of control and video links. A quiet drone still moves air, and still talks to its pilot.
Cross-cue and friend or foe
Contacts combine across the mesh. A cooperative drone broadcasts Remote ID. Its absence around a track is itself a signal.
Triangulate, honestly
Enough bearings give a fix with a clear uncertainty ellipse. Fewer nodes give a coarser fix, and the system says so.
Commodity parts. Emits nothing.
Built entirely from commercially available parts. Affordable to field at scale, and built to be attritable. It only listens, and never transmits.
| Compute | Quad-core ARM64 single-board computer (Linux) |
| Acoustic | USB or I2S MEMS microphone array |
| RF | SDR receiver, 2.4 and 5.8 GHz |
| Remote ID | ESP32 OpenDroneID receiver |
| Timing and position | u-blox GNSS (time and node fix) |
| Emission | None. Receive only, no active radar. |
| BOM target | About €150 per node |
| Power | Battery or solar capable, low draw |
Scales with nodes. Fails gracefully.
- / /Coverage scales linearly. More nodes, more overlap, tighter fixes.
- / /Nodes cue each other. A weak hit becomes a confident track when neighbors agree.
- / /No single point of blindness. Lose a node and the mesh reroutes. Resolution drops, awareness does not.
Systems that overclaim get people hurt.
Detection is probabilistic and bounded by physics. We treat that as engineering, not marketing.
- 01Every track carries a confidence.
- 02Uncertainty is shown, never hidden.
- 03We state where the sensing ends.
New sensors plug into one shared schema.
Each new layer publishes to the same fusion schema the acoustic and RF layers already use.
Ground vibration
Geophone sensing for low, close passes.
Magnetic anomaly
Motor and battery signatures as a corroborating cue.
Passive infrared
Low-power thermal triggers that wake the richer sensors.
Visual cueing
Slew to cue optical confirmation on an existing track.
Talk to us about the uncooperative airspace.
Built for the NATO DIANA application. For a briefing or evaluation, get in touch.
Triangle Mesh is an independent effort. This page uses no NATO or DIANA marks and implies no endorsement, affiliation, or selection. All diagrams are our own technical illustrations, not screenshots of live detections.