ATLAS Gunshot · TAK application

Locate gunshots hands-free

ATLAS Gunshot Localization connects GPS-ready phones to a gunshot-processing server. Together, participating devices detect an event and turn its arrival timing into a location on the TAK map.

Talk to us ↗

ATAK map showing a small gray dot titled Gunshot Detected — When a gunshot is detected, ATAK displays a small gray map marker titled “Gunshot Detected.”

InterfaceOne-page TAK front end

MethodTime difference of arrival

NetworkCollaborative phone sensing

01 — Simple to join

Record when your device is ready.

The app keeps the front end focused on one action. When the device has enough GPS satellites, the user can tap Record to connect to the gunshot-processing server and begin contributing observations.

As more phones participate, the system has more arrival-time measurements to work with when a gunshot occurs.

01Confirm GPS readiness
02Tap Record to join
03Share event timing with the processing server

ATLAS Gunshot Localization main recording screen — A single screen for checking GPS readiness and starting a recording session.

02 — TDOA localization

Arrival-time differences draw the path to a solution.

Each device observes a shot at a slightly different time. Those differences generate hyperbolas on the map; their intersection produces a gunshot localization.

Anti-sniper concept showing gunshot localization hyperbolas on a map — Anti-pop-shot concept: a small group of phones contributes hyperbolas.

Anti-sniper concept showing a gunshot localization from several phones — Even a few participating phones can produce a useful localization.

03 — More sensors, more evidence

Dense participation creates a stronger geometric picture.

With many connected phones, more hyperbolas contribute to the map and support a well-constrained localization solution.

Map showing many time-difference-of-arrival hyperbolas — Multiple observations produce a dense set of TDOA hyperbolas.

Map showing gunshot localization generated from many hyperbolas — A larger participating network yields a clear localization solution.

04 — Machine-learning model

Machine learning for reliable muzzle-blast timing.

ATLAS identifies the acoustic arrival that matters for localization, even when a shockwave arrives first.

ATLAS uses a machine-learning model to distinguish the muzzle-blast sound from a ballistic shockwave when one is present. For a down-range phone and a supersonic projectile, the shockwave can arrive before the muzzle blast; the vertical marker in each example shows the model’s muzzle-blast estimate for the localization workflow.

Gunshot audio example without a shockwave, with the model's muzzle blast marker — No shockwave: the model marks the muzzle-blast arrival.

Gunshot audio example with a preceding shockwave and the model's muzzle blast marker — Shockwave present: the model distinguishes it from the later muzzle blast.

User view

A concise alert, right on the map.

From the user’s perspective, the detected event appears as a small gray point on the ATAK map titled “Gunshot Detected.”

ATAK map with a small gray dot titled Gunshot Detected — The ATLAS Gunshot user-facing event marker in ATAK.

Next project
ATLAS SIGINT↗