AxiomLabs × IISER Tirupati Hackathon

Mission-Grade
Orbital Intelligence

Real-time satellite tracking, AI-powered collision avoidance, and orbital mechanics — all in a single mission-control dashboard.

Watch Demo View Source
3D
Earth Visualization
7
Analytics Modules
5
AI Reasoning Steps
2s
Telemetry Refresh
Demo

See It In Action

A full walkthrough of the dashboard — from 3D orbital visualization to the AI conjunction analysis pipeline.

🛰 3D Earth + Satellite Fleet
🤖 AI Reasoning Panel
📡 Live Telemetry Bar
⚠️ Conjunction Alerts
🗺 Ground Track Map
Capabilities

Built for the Real Problem

Every feature is aimed at the hardest part of satellite operations: knowing when two objects are about to meet in the worst way possible.

🌍
Immersive 3D Globe
Full Three.js-rendered Earth with live satellite positions, orbital trails, debris fields, and risk heatmaps. Rotate, zoom, inspect any object in the fleet.
🤖
Live AI Reasoning Panel
The AI doesn't just alert — it explains. Watch it walk through its math step-by-step: Tsiolkovsky equations, TCA calculations, and delta-v planning in real-time.
⚠️
Conjunction Detection
Orbital mechanics backend continuously computes Time of Closest Approach (TCA) and probability of collision. Miss distance and Pc thresholds trigger escalating alerts.
🛸
Maneuver Planning
Calculates optimal avoidance burns using Hohmann transfers. Fuel consumption, delta-v budget, and post-maneuver trajectory all visualized in the Maneuver module.
🗺️
Ground Track Prediction
2D Mercator projection showing sub-satellite point paths, ground station coverage cones, and predictive pass windows over any ground location.
📊
Pre-Path CDM Reports
Full Conjunction Data Messages in the analytics hub, modeled after NASA/CARA format. View historical conjunction records with resolved/active status tracking.
AI System

The Reasoning Engine

The AI panel runs a 5-phase cognitive loop on live telemetry data — not a static display, but active computation per simulation tick.

01
SCAN
Ingests all active satellite TLEs and computes current state vectors from the FastAPI snapshot endpoint.
02
PREDICT
Propagates orbits forward using SGP4 physics. Screens all object pairs for potential conjunction events within the time window.
03
EVALUATE
Computes Pc (probability of collision), miss distance, and TCA for flagged pairs. Applies Tsiolkovsky to assess maneuver feasibility.
04
PLAN
Generates optimal avoidance maneuver — burn timing, delta-v magnitude, and direction. Validates against propellant budget constraints.
05
EXECUTE
Dispatches maneuver recommendation with confidence score. Updates simulation state and reruns SCAN for verification.
// The math shown in the AI panel is live — Tsiolkovsky equation, Pc computations, and delta-v values are pulled from the backend /snapshot endpoint on every 2-second poll cycle.
2D Hub

7 Intelligence Modules

Switch from the 3D globe into a mission-critical analytics hub with seven dedicated intelligence pages.

01
Fleet Overview
Top-level fleet status: active satellites, health scores, anomaly flags, and orbital regime distribution.
02
Ground Track
Live 2D ground tracks, coverage zones, and predictive pass windows over any point on Earth.
03
Conjunction Analysis
Bullseye risk charts, Pc timelines, and ranked conjunction event queue with severity codes.
04
Maneuver & Fuel
Delta-v budget tracker, burn history, fuel gauge per satellite, and maneuver timeline playback.
05
System Performance
Backend latency, telemetry frame rates, physics computation load, and simulation health metrics.
06
Pre-Path CDMs
Conjunction Data Messages modeled on CARA/NASA format. Full history of close approach events with resolution status.
07
AI Reasoning Hub
Deep-dive into the AI's current decision chain. Step-by-step math, confidence breakdowns, and recommendation rationale.
Architecture

The Stack

Chosen for speed, visual power, and hackathon pragmatism — no unnecessary abstractions.

Frontend
React + Vite
Blazing-fast dev server on port 5173. JSX components with direct CSS — no Tailwind, no component libraries. Absolute control over every animation.
Port 5173
3D Engine
Three.js
Custom SceneManager, Earth mesh with atmosphere shader, satellite instancing, debris particle system, and risk heatmap overlay. OrbitControls for free camera.
WebGL
Backend
FastAPI
Python 3.9+ with Uvicorn on port 8000. Orbital mechanics via SGP4 propagation. Snapshot endpoint serves live telemetry JSON every poll cycle.
Port 8000
Physics
Orbital Mech
Custom physics.py handles SGP4 propagation, Hohmann transfers, Tsiolkovsky Δv equations, TCA computation, and Pc estimation for conjunction events.
physics.py
Styling
Raw CSS
~49,000 lines of hand-crafted CSS. All glowing dots, pulsing risks, and animated metric bars are pure CSS — zero JS animation libraries.
index.css
Transport
Polling API
2-second REST polling instead of WebSockets. Fast enough for orbital time scales — objects move fast, but they're very far away.
telemetryClient.js
Setup

Get It Running

Two startup scripts handle everything automatically. Or go manual if you prefer to control each step.

▸ Quick Start (Windows)
# Double-click or run in terminal
start.bat

# Opens backend + frontend
# in separate windows automatically
▸ Quick Start (Mac / Linux)
bash start.sh
# or
./start.sh
▸ Manual Setup
# 1. Backend
cd backend
python -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt
uvicorn main:app --reload --port 8000

# 2. Frontend (new terminal)
cd ..
npm install
npm run dev

# Open browser at:
http://localhost:5173
⚠  Requirements:  Python 3.9+  ·  Node 18+  ·  pip  ·  npm
   If CORS errors appear, verify Uvicorn is bound to 0.0.0.0 and port 8000 matches the frontend config.
Team

AxiomLabs

Hackathon Team — IISER Tirupati
Indian Institute of Science Education and Research, Tirupati
Orbital Mechanics
3D Visualization
AI Systems
Mission Control UI
Space Safety
BUILD TIME
~48h
incl. 4AM CSS sessions
"Huge shoutout to the team for grinding on the Three.js optimizations! We literally stayed up until 4 AM fixing the CSS animations, but it was worth it." — AxiomLabs README