Offline search and rescue personnel tracking over mesh radio networks.
SARMesh listens to a Meshtastic mesh, records position beacons from field trackers, and plots them against the teams working an incident. It runs entirely off-grid: no internet, no cloud, no cell coverage. Everything lives in a single local SQLite file.
A tracker is any Meshtastic node carried by field personnel. Nodes broadcast position packets over LoRa, and a node connected by USB to the machine running SARMesh acts as the receiver.
field trackers ──LoRa──> base node ──USB──> SARMesh ──> sarmesh.db
Each position is stamped with the incident its tracker is assigned to, so one mesh can serve consecutive operations without the logs running together.
Status: early development. The ingest pipeline, map UI, incident/team/ tracker management, offline basemaps and track history all work end to end. Trackers that stop beaconing are shown as stale, but nothing alerts on it yet, and GPX export is not built.
This is a situational awareness aid, not a life-safety system. Positions arrive over a lossy radio mesh and can be minutes stale or missing entirely. Do not use SARMesh as the only means of tracking personnel in the field.
- Python 3.12+ and uv
- A Meshtastic device on USB serial, plus one or more tracker nodes on the same channel
PySide6 ships prebuilt wheels for Linux (x86_64 and arm64), Windows and macOS, so there is nothing to compile. On a Raspberry Pi you need Pi OS Trixie or newer, because the arm64 wheels want glibc 2.39 and Bookworm ships 2.36.
uv sync
uv run sarmesh appThat opens the desktop app: a Python backend serving a React map UI inside a Qt window. Qt bundles its own rendering engine, so there is no dependency on a system browser or on distro packages like WebKitGTK.
uv run sarmesh app --basemap terrain.mbtiles # with an offline basemap
uv run sarmesh app --host 192.168.1.50 # reach the radio over the network
uv run sarmesh app --offline # stored data only, no radio
uv run sarmesh app --browser # serve only, no native windowIf no radio is reachable the UI still starts, so an operator can read the last known positions and manage teams without one. The server also answers on the local network, so a second device can view the incident in a browser.
Create an incident, add teams and trackers, and assign trackers from the UI. The same operations are available from the command line: see docs/operations.md.
Run from a source checkout, everything writes to sarmesh.db in the working
directory. A packaged build is launched by double-clicking and cannot rely on
that, so it uses the per-user data directory:
| Platform | Location |
|---|---|
| Linux | $XDG_DATA_HOME/sarmesh/ (default ~/.local/share/sarmesh/) |
| macOS | ~/Library/Application Support/SARMesh/ |
| Windows | %LOCALAPPDATA%\SARMesh\ |
Set SARMESH_DB to override both, which is how you point a build at an incident
database on removable media. Every command reads it, so the CLI and the app
always agree on which database they are using.
| Operations | Running an incident, the CLI, logs, offline basemaps |
| Architecture | Data model, layout, how the layers fit together |
| Development | Dev setup, the check commands, working on the frontend |
| Packaging | Building the desktop bundle for each platform |
Contributions are welcome. Start with docs/development.md for the dev loop, and docs/architecture.md for where code belongs.
Copyright (C) 2026 Preston Roesslet
SARMesh is free software under the GNU General Public License v3.0,
version 3 only. It links the meshtastic library, which is GPL-3.0-only, so
the combined work cannot be offered under a later version.
This program is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the license for details.
Built on Meshtastic, whose LoRa mesh firmware and Python library do the hard part. Maps are rendered with Leaflet; offline basemaps use the MBTiles format.