Hey all,
I’ve been building a tool called Tarmac and wanted to get feedback from the people who’d actually use it before going further.
Where it came from:
We’ve been doing ArduPilot companion-side development internally for a while — some of it open source, some of it not yet. Tarmac is the hosted, point-and-click version of that work: operators who don’t want to self-host get up and running quickly, and the revenue funds continued open source development so improvements flow both ways.
What it does:
A lightweight Python service runs on the companion Pi, connects to ArduPilot over MAVLink via MAVSDK, and streams a full telemetry dump — attitude, position, GPS, battery, IMU, RC status, actuators, flight mode — to a web dashboard in real time. A second service handles a job queue so you can push missions, pull logs, change parameters, or trigger OTA firmware updates from a browser without needing to be on the same network.
The part I think is actually interesting — the App Builder:
The bigger idea is making custom payload integration accessible to people who aren’t backend developers. Right now if you’re flying something non-standard — a custom sensor, a smart battery, a gimbal, a lidar, a gas detector, whatever — getting that data into a usable UI typically means writing your own ingestion pipeline, your own storage, your own frontend. That’s a significant lift for most operators and it means a lot of good integration work never gets shared.
The App Builder is an attempt to close that gap. You define a data schema and a lightweight parser function for your payload, and Tarmac generates the REST ingestion endpoint, handles storage, and builds a live UI widget from it automatically. The companion-side script is a single Python file. The goal is that integrating a new payload goes from a multi-week project to an afternoon — and that those integrations can be shared across the community rather than everyone rebuilding the same thing independently.
What I’d genuinely like to know:
-
Is fragmented tooling a real pain point, or is QGC/MP enough for your workflow?
-
For anyone running more than one vehicle — what does your current ops stack look like?
-
What payloads are you flying that have no good integration story right now?
-
Would a sharable payload integration ecosystem actually be useful, or is everyone’s setup too bespoke for that to matter?
-
Hosted vs self-hosted — does it matter to you?
Still building, happy to be told there’s no gap here if that’s the honest answer.
Thanks