ISRO Chandrayaan-5: Generative Agents Simulation

A multi-agent simulation of ISRO's first permanent lunar base, implementing Stanford's Generative Agents architecture (2023) with 8 AI agents that develop emergent social behaviors.

🚀 Project Overview

Setting: Aryabhata Station - ISRO's lunar base at Moon's South Pole (Year 2035)

Architecture: PARL Framework (Perception, Action, Reasoning, Learning) with Stanford-style features.

👨‍🚀 The Crew (8 Agents)

Agent	Role	Key Traits
Cdr. Vikram Sharma	Mission Commander	Disciplined, decisive
Dr. Ananya Iyer	Botanist/Life Support	Nurturing, optimistic
TARA	AI Assistant	Curious, logical, evolving
Priya Nair	Crew Welfare Officer	Empathetic, perceptive
Aditya Reddy	Systems Engineer	Practical, homesick
Dr. Arjun Menon	Flight Surgeon	Calm, analytical
Kabir Saxena	Geologist/Mining Lead	Rebellious, brilliant
Rohan Pillai	Communications Officer	Cheerful, anxious

🧠 Stanford Features Implemented

Feature	Description
Memory Stream	FAISS-based with importance scoring and decay
Bidirectional Memory	Both speaker AND listener remember conversations
Information Propagation	Track who told whom what
Relationship Tracking	0-100 strength with sentiment
Daily Planning	Role-based schedules
Triggerable Events	Inject information and watch it spread
Analytics	Track emergent behavior propagation

🏗️ Tech Stack

Layer	Technology
Frontend	React + Vite + Three.js
Backend	Python + FastAPI
Real-time	WebSocket
Primary LLM	Groq API (llama-3.1-8b-instant)
Fallback LLM	Ollama (local)
Memory Store	FAISS (vector DB)

🚀 Quick Start

Backend

cd backend
python -m venv isro_env
source isro_env/bin/activate
pip install -r requirements.txt
# Add GROQ_API_KEY to .env
./run.sh

Frontend

cd frontend
npm install
npm run dev

Open http://localhost:5173 and click "Start Simulation"!

📡 API Endpoints

Agent Details

• GET /api/agents/{name}/memories - Memory stream
• GET /api/agents/{name}/relationships - Relationship data
• GET /api/agents/{name}/plan - Daily schedule
• GET /api/agents/{name}/full - Complete info

Events & Analytics

• GET /api/events - List available demo events
• POST /api/events/{id}/trigger - Trigger an event
• GET /api/analytics - Propagation summary
• GET /api/analytics/event/{id} - Event spread analysis

🎯 Demo: Emergent Behavior

1. Start simulation with 8 agents
2. Trigger an event:

   curl -X POST http://localhost:8000/api/events/crew_meeting/trigger

3. Watch agents spread information through conversations
4. Check who knows about the meeting:

   curl http://localhost:8000/api/analytics/event/crew_meeting

📁 Project Structure

IntelligentAgent/
├── backend/
│   ├── app/
│   │   ├── agents/
│   │   │   ├── generative_agent.py  # 8 agent definitions
│   │   │   └── relationships.py     # Relationship tracking
│   │   ├── parl/
│   │   │   ├── parl_engine.py       # LLM reasoning
│   │   │   └── planner.py           # Daily schedules
│   │   ├── simulation/
│   │   │   ├── engine.py            # Main loop
│   │   │   ├── events.py            # Demo events
│   │   │   └── analytics.py         # Propagation tracking
│   │   └── memory/
│   │       └── memory_store.py      # FAISS memory
│   └── requirements.txt
│
├── frontend/
│   └── src/
│       └── components/
│           ├── LunarBase.jsx        # 3D visualization
│           └── AgentPanel.jsx       # Agent details panel
│
├── README.md
└── D1_Agent_Design_Document.md

📄 Reference

Based on: Park, J.S., et al. (2023). "Generative Agents: Interactive Simulacra of Human Behavior." arXiv:2304.03442