Genesis-Embodied-AI/genesis-world

Health v2 · maintenance only

Adoption not part of the score

29809 stars · 2844 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-29, confidence not recorded

Genesis World is a Python simulation platform for physical AI and robotics that combines a unified multi-physics engine, a photorealistic renderer (Nyx), and a cross-platform compiler (Quadrants) behind a single Pythonic API. It scales from laptop CPUs to datacenter-grade GPUs and supports asset formats like URDF, MJCF, USD, and GLB.

Use cases

• simulate robots for embodied AI research
• train reinforcement learning policies in physics simulation
• generate synthetic training data for robotics
• render photorealistic sensor data from simulated scenes
• run parallel physics environments on GPUs
• simulate soft bodies, fluids, and rigid bodies in one scene

When to choose

• you need a unified multi-physics engine for robotics research
• you want GPU-accelerated simulation that scales to datacenter hardware
• you need photorealistic rendering integrated with physics simulation
• you prefer a Pythonic API embeddable in ML pipelines

When to avoid

• you need a battle-tested simulator for safety-critical industrial deployment
• you require a lightweight 2D physics engine for simple games
• you need a mature technical report or long-term stability guarantees, as the project is young and evolving

Facets

library · maturity active

simulation graphics machine-learning robotics simulation robotics machine-learning artificial-intelligence gpu-computing python cross-platform windows physics-engine embodied-ai robotics-simulation multi-physics renderer reinforcement-learning-environments urdf usd gpu linux macos

Member repositories

For agents

markdown · JSON · MCP: product_card(name="Genesis-Embodied-AI/genesis-world")