Long exposure photographs with automated Ardupilot Quad(s).

This project uses uv for Python package management and includes MAVSDK-Python as a git submodule for better agent/developer experience.

# Clone with submodules
git clone --recursive <repo-url>

# Or if already cloned, initialize submodules
git submodule update --init --recursive

# Install dependencies with uv
uv sync

The MAVSDK-Python library is included as a submodule at external/mavsdk-python/ for:

• 🔍 Source code inspection - Useful for AI agents and developers
• 📝 Proto definitions - Available at external/mavsdk-python/proto/protos/
• 🛠️ Development - Make changes and contribute back to MAVSDK
• 📚 Documentation - Inline with the code

# Inspect available MAVSDK API methods
uv run python inspect_mavsdk_api.py

# View telemetry proto definitions
cat external/mavsdk-python/proto/protos/telemetry/telemetry.proto

# Update MAVSDK-Python to latest
cd external/mavsdk-python
git pull origin main
cd ../..
uv sync

• Proof of concept / light weight prototype of long exposure drone photography tool
• Give a interface to upload a SVG (or similar) path and have a quadcopter fly said path in the flight region

• Run on a single laptop
• Ardupilot runs on quad w/ remote telementry link to laptop running main stack
• Data logging / viz for debugging
• SIL testing
• User can upload a SVG path and have a quadcopter fly said path in the flight region
• External control method for ground station like control

• Stack:
  • Python / UV based prototype
  • Rerun for dataviz
  • Ardupilot based quadcopter w/ mavlink for comms
  • Single / near-single application for all purposes
    • Sub tasks
  • Docker for wrapping
• Remote Control:
  • HTTP API for requesting data such as ardupilot fields / parameters
  • HTTP API for sending commands such as ARM or our high level control commands
  • CLI based control app / TUI (bonus / optional) using said HTTP API for Stage 1 demo.
    • Vibe coded GCS using API spec ok for this version
• Main Application:
  • Python application running tassk such as the main ardupilot loop, our web API, our custom logic
  • Shared state (thread locked ideally) of the quad
    • Can either choose a shared mutex state or a message queue / table based approach
    • Probably shared mutex state is easier for this version
  • UPDATE: Message based.
    • Best matches ardupilot
    • Rerun can log to convert to time series
    • All tasks really only care about the message and not the state
    • Can just be an array per task in the task manager w/ sub system
    • This allows for easier testing and debugging
    • Commands are now easier to add in
      • tasks just listen for when a command is not null and runs.
    • In theory nothing is stateful. (tasks can store their own state or we can at least store last values)
    • Maybe a PUB/SUB/GET/SET Model?
• Logging:
  • Embedded as a task is a Rerun based loggering logging at every tick (or event)
• Autpilot and link
  • Ardupilot based quadcopter
  • Wifi for development but idelaly 900mhz or ELRS based telementry link for test.
  • Connects to laptop over mavlink.
  • Can SIL with docker container w/ full stack for accurate testing
• Camera controll
  • Need a way to actually do lon exposed
  • Amazon a controller
  • Design a tool + PCB to do this if im cool
  • RP2350 time?

• Setup base application w/ task loop
• Setup Ardupulot connection w/ SIL Container
• Setup heartbeat task for Ardupilot
• Setup rerun w/ hello world logging of ardupilot data
• SIL Hop w/ Viz
• Move logging to own class (reached file length limit)
• LED Control w/ Viz (mock)]
• 3D Viz
• Long Exposure viz (2d and 3d)
• Follow Waypoint path
• Track color in waypoints
• Add Low Poly Mesh desert floor / grid
• Add Solid paths (waypoints with solid color)
• Add Wait for takeoff instead of time based
• Increase waypoint accuracy (hold point improvement)
• Auto stop after landing
• SVG Importer
• 3D Path Importer (blender sourced)

• Render pointclouds / depth clouds using 2.5d - draw points along depth.

• LED Control
  • Optional Replacement: Develop Mavlink based LED Control
  • Develop ESP32 based LED Controller that uses WiFi or 900mhz link
  • Develop Connection to LED Controller w/ hook up to software LED component
• Rust Re-write
  • Basic project setup
  • Ardupilot connection w/ POC state reading / Status message logger
  • MVP Rerun logger w/ Status Messages (Text Log)
  • Port over all UAV state reading
    • Local Position NED w/ History
    • Lattitude, Longitude, Altitude w/ History
    • Health
  • Port LED Software Class
    • Basic class and creation
    • LED control commands
    • LED viz
  • Port over all Rerun Logging
    • GPS Map View (LLA)
    • 3D NED Live View
    • 3D Historical / Exposure View
    • 2D Historical / Exposure View
  • Port over pattern API w/ generators
  • Port over waypoint system
  • Port over mission system
  • Docker compose SIL setup / tear down
• Clean Up Pass #1
  • Remove Seperate tasks - instead make direct call backs
  • Send outgoing mavlink in the main task loop
  • Remove App layer?
• Flight Test Campaign
  • Arm/Disarm Test - Arm the quad - set LEDs - disarm
  • Motor Spin up test
  • Hop test
  • Single Waypoint test
  • Multi-waypoint test
  • Square pattern test
  • Smiley pattern test

• Track System / Sensor Status from MavLink
  • https://github.com/victoryforphil/project-firefly/blob/main/src_flight/whisper/src/mavlink/processors/proc_sys_status.rs
• Add App logging to Rerun
• UAV Mesh to Viz