OrbitGuard

OrbitGuard is a public-data powered satellite conjunction screening, collision probability analysis, and avoidance planning platform.

Capabilities

• TLE Ingestion: Automatic ingestion from CelesTrak (every 6 hours) with manual refresh
• SGP4 Propagation: TLE-based orbit propagation over configurable windows
• Cowell Numerical Propagation: High-fidelity propagation with J2-J6 harmonics, drag, SRP perturbations
• TCA Root-Finding: Sub-second precision closest approach detection using Brent's method
• B-Plane Analysis: Full B-plane geometry (BdotT, BdotR) per CCSDS 508.0-B-1
• Probability of Collision: Three methods — Foster (2D Gaussian), Chan (maximum Pc), Monte Carlo with confidence intervals
• Covariance Handling: TLE-based estimation, positive-definite validation, Higham correction
• Maneuver Planning: RTN frame impulsive burns, minimum delta-v optimizer, 6-direction trade study
• Fuel Cost: Tsiolkovsky rocket equation with configurable Isp
• Risk Classification: NASA CARA operational thresholds (GREEN/YELLOW/RED)
• 3D Visualization: Three.js-based event replay with pre/post-maneuver toggle
• Congestion Analysis: Altitude-band density and conjunction rate metrics

Repo layout

• backend/ — FastAPI app with orbital mechanics core
• frontend/ — Next.js app with 3D visualization
• docs/PHYSICS.md — Mathematical foundations and equations
• docs/VALIDATION.md — Test results against reference implementations
• prd.md — Product requirements

Backend setup

1. Create a virtual environment and install dependencies:

cd backend
python -m venv .venv
source .venv/bin/activate  # or .venv\Scripts\activate on Windows
pip install -r requirements.txt

2. Verify dependencies:

python verify_dependencies.py

3. Run tests:

python -m pytest tests/ -v

4. Run API:

uvicorn app.main:app --reload --host 0.0.0.0 --port 8000

API Endpoints

Method	Path	Description
GET	/catalog	List tracked satellites
POST	/catalog/satellite	Add satellite by NORAD ID
GET	/catalog/positions	Current positions for all tracked objects
GET	/assets/{norad_id}/conjunctions	Find conjunctions for a defended asset
GET	/conjunctions/{event_id}	Full conjunction event detail (B-plane, Pc, covariance)
POST	/conjunctions/{event_id}/avoidance-sim	Simulate an avoidance maneuver
POST	/events/{event_id}/optimize-maneuver	Find minimum delta-v for target Pc
GET	/events/{event_id}/trade-study	6-direction maneuver trade table
GET	/congestion	Altitude congestion metrics
POST	/ingest/refresh	Manual TLE refresh

Frontend setup

cd frontend
npm install
npm run dev

Units Convention

• Distances: km
• Velocities: km/s
• Delta-v: m/s
• Probability: raw float (0-1)
• Mass: kg
• Time: ISO 8601 UTC

Architecture

backend/app/services/
├── orbital_constants.py    # Physical constants, PropagationConfig, RTN frame
├── cowell_propagator.py    # Numerical integration with perturbations
├── tca_finder.py           # Root-finding for closest approach
├── bplane_analysis.py      # B-plane geometry computation
├── covariance_handler.py   # Covariance estimation, validation, projection
├── pc_calculator.py        # Foster, Chan, and Monte Carlo Pc methods
├── maneuver_planner.py     # RTN burns, optimizer, trade study, fuel cost
├── avoidance_simulator.py  # API-facing simulator (wraps maneuver_planner)
├── screening_engine.py     # Full pipeline: propagate → TCA → B-plane → Pc
├── propagate_engine.py     # SGP4 propagation
├── ingest_service.py       # TLE ingestion from CelesTrak
└── congestion_analyser.py  # Altitude band congestion metrics