pymavrest

pymavrest is a high-performance REST server for MAVLink-based drones 🚁. It lets you read telemetry over HTTP and send flight commands over HTTP.

It is a Python alternative to mavlink2rest with a strong async architecture 🛰️: https://github.com/mavlink/mavlink2rest

Project strengths ✨

• Fully async and high performance ⚡.
• Supports VTOL, multirotor, and other MAVLink-capable vehicles 🚁.
• Compatible with PX4, ArduPilot, and other MAVLink firmwares 🛩️.
• Read telemetry data via HTTP and send drone commands via HTTP 🌐.
• Built-in auth and multi-user access control (JWT + user permissions) 🔐.
• Uses pymavlink and mavsdk backends and can switch automatically when needed 🔄.
• Highly configurable via config.json ⚙️.
• Easy one-command install on mini-computers like Raspberry Pi 🖥️.

Production prerequisite 🛫

For production deployments, install and configure mavlink-router:

https://github.com/mavlink-router/mavlink-router

Data flow and API example 📡

Architecture:

Drone 🚁 <-> mavlink-router 📡 <-> pymavrest 🛰️ <-> REST Clients 🌐

Example request (flight_details):

curl http://localhost:8080/api/v1/messages/flight_details \
  -H "Authorization: Bearer <token>"

Returns aggregated telemetry including:

• flight mode 🧭
• battery state 🔋
• GPS position 📍
• speed and attitude 🏎️
• RC status 🎮
• EKF status 📈
• connection flags 🔗
• mission summary 🗺️

Note: use the port configured in config.json (rest_api.port). The default in this project is 10821.

OpenAPI docs are available at 📘:

http://127.0.0.1:10821/docs

The port may differ depending on your config.json (rest_api.port).

Install (one command) ⚙️

Run installer directly with curl:

curl -fsSL https://raw.githubusercontent.com/hadif1999/pymavrest/master/scripts/install.sh | bash -s -- --prod true --backend pymavlink

This installs dependencies, syncs the repo into ~/app by default, and in production mode deploys/runs pymavrest.service.

Installer arguments 🧩

scripts/install.sh supports:

• --branch <name>: Git branch to clone/pull. Default: master
• --app-dir <path>: Install/update directory. Default: $HOME/app
• --prod <true|false>:
  • true: deploy and run with systemd
  • false: install only (do not start app)
  • Default: true
• --backend <pymavlink|mavsdk>: Backend passed to runtime. Default: pymavlink
• -h, --help: show full help

Help example:

curl -fsSL https://raw.githubusercontent.com/hadif1999/pymavrest/master/scripts/install.sh | bash -s -- --help

Configure config.json (important for production) 🛠️

Before production use, edit your config.json values for your environment. At minimum, update drone connection, auth secrets/users, server endpoints, and networking.

If running as a service, restart after config changes:

sudo systemctl restart pymavrest

config.json parameters 📘

See README_config.md for full details. Main sections:

• general
  • log_level: logging level (DEBUG, INFO, ...)
  • is_production: production behavior flag
• drone.properties
  • URI: MAVLink source (example: tcp://127.0.0.1:14550)
  • FLIGHT_MODES: mode code mapping
  • battery_cap_mah: battery capacity metadata
  • serial_number: optional drone identifier
• external_devices
  • Device configs such as gps and sms (type, COM, baud, enabled)
• requests
  • ping_check_by_host, timeout, retries
• rest_api
  • port, host, global_prefix, global_timeout
  • as_https, ssl_keyfile_dir, ssl_certfile_dir
• auth
  • enabled, jwt_secret, jwt_token_expire_minutes, jwt_algorithm
  • users[] with username, password, permission, active, is_admin
• health_check
  • enabled, push_to_server, send_sms_in_fail
  • route, flight_info_route, update_interval_sec
• server
  • base_url
  • auth: username, password, route, type
• sms
  • recipient

Run manually (test or production) ▶️

You can run directly with uv run main.py for testing or production-like manual operation:

cd ~/app
uv run -p 3.12 main.py -c config.json --backend pymavlink

Or with mavsdk:

uv run -p 3.12 main.py -c config.json --backend mavsdk

For long-running production systems, prefer systemd mode (--prod true) so service restarts are managed automatically.