🌍 WorldSim AI — Digital Twin Simulation Platform
Model, predict and optimize real-world systems. Agent-based simulation with real-time 3D visualization, machine-learning forecasting, reinforcement learning and distributed scaling — free and open source (MIT), one Docker command to run.
- 🌐 Project page: https://rudra496.github.io/worldsim-ai/
- 💻 Source: https://github.com/rudra496/worldsim-ai
- 📄 DOI: 10.5281/zenodo.22118796
What it does
- Agent-based modeling — Vehicle, Human, Machine and EnergyUnit agents in rule-based or probabilistic worlds; Planner/Predictor/Optimizer agents coordinate via an AgentCoordinator
- 8 zone types — residential, industrial, commercial, road, park, power plant, water treatment, warehouse — with resource production/consumption tracking
- AI layer — PyTorch LSTM forecasting, autoencoder anomaly detection, scipy LP resource optimization, Gymnasium-compatible RL (PPO / Q-learning) closing the simulation → prediction → optimization loop
- Real-time 3D — React Three Fiber world with orbit/pan/zoom, day-night cycle and 2D ↔ 3D switching
- Live data ingestion — MQTT (IoT sensors), CSV/JSON/JSONL, REST polling, or synthetic sources with noise, drift and failure injection
- Distributed + digital twin — multi-node gRPC engine with spatial partitioning and load balancing; live/replay/hybrid sync, GeoJSON + geofencing, plugin system (Prometheus, Slack)
One-command start
git clone https://github.com/rudra496/worldsim-ai && cd worldsim-ai
docker-compose up --build # open http://localhost:3000
For
Smart-city modelers, factory-floor optimizers, energy-grid researchers and anyone who needs a digital twin without enterprise licensing.
Tech stack
Python 3.10+ · FastAPI · React 18 · Three.js · NumPy · optional PyTorch · gRPC · Docker
Log in or sign up for Devpost to join the conversation.