🧭 PNT Dead Reckoning

PNT (Position, Navigation, and Timing) Dead Reckoning is a robust sensor fusion pipeline designed for accurate navigation in GPS-denied environments. The system relies on a Python/OpenCV visual odometry provider and an IMU driver to gather raw spatial data. A high-performance Go-based navigation core ingests this data, applying a Square-Root Information Filter (SRIF) to calculate coordinates, velocity, and orientation. Operators can monitor the calculated path, visual tracking features, and drift uncertainty in real-time via a React/TypeScript tactical dashboard.

• Features
• Technologies Used
• Installation
• Usage
• Project Structure
• Contributing
• License

• Visual Odometry (VO): A Python application utilizing OpenCV to detect ORB/SIFT keypoints and calculate optical flow, streaming velocity vectors via Unix sockets.
• SRIF Sensor Fusion Brain: A Go-based navigation core that reads raw IMU (Accel/Gyro) data and fuses it with the VO feed using an advanced Square-Root Information Filter (SRIF) to maintain a highly accurate 9-DOF state vector.
• Tactical Navigation HUD: A React and TypeScript interface featuring a Mapbox/Leaflet live breadcrumb trail, real-time CV feature tracking overlays, and a visual drift indicator (Uncertainty Ellipse).
• Local Simulation: Includes Python scripts to generate "True" versus "Noisy" drone path data, allowing for rapid local testing of the filter's stability without physical hardware.

• Navigation Core: Go
• Vision Provider: Python, OpenCV, GStreamer
• Frontend Dashboard: React, TypeScript, Mapbox/Leaflet
• Inter-Process Communication: Unix Sockets, WebSockets
• Infrastructure: Docker, Docker Compose

1. Clone the repository:

   git clone https://github.com/yourusername/pnt-dead-reckoning.git
   cd pnt-dead-reckoning

2. Boot the full stack (Vision Provider, Navigation Core, and Tactical UI) using Docker Compose:

   docker-compose up -d

3. Optional: If you do not have camera hardware attached, run the local simulator to generate sample telemetry:

   python simulation/drone_path_generator.py

• Access the Tactical Map: Open your browser and navigate to http://localhost:3000 (or your configured port) to view the Live Breadcrumbs and Drift Indicator.
• Monitor the Vision Feed: Toggle the Vision Feed overlay in the UI to see real-time optical flow and keypoint extraction processing from the camera.
• Analyze State History: The Go daemon serves coordinate history over WebSockets, allowing the UI to instantly trace the historical path of the agent.

• /vision_provider: Python/OpenCV container handling raw frame capture and visual odometry calculations.
• /navigation_core: Go application executing the primary SRIF math and state vector management.
• /tactical_map: React frontend providing the real-time PNT mapping interface.
• /simulation: Tools for generating noisy test data for filter calibration.
• /.github/workflows: CI/CD pipelines including rigorous validation for the SRIF matrix math.

Contributions are highly encouraged! When submitting a Pull Request, especially involving the /filter directory, please ensure all automated tests for filter stability and SRIF matrix math pass successfully.

This project is licensed under the MIT License - see the LICENSE file for details.