Gibson Environment

Gibson Environment is a pioneering photorealistic simulation environment for embodied AI developed at Stanford University. It introduced the concept of virtual-to-real transfer for embodied agents, providing 1,400+ real-world 3D indoor space reconstructions scanned from real buildings. Gibson provides a perception-based simulation engine that renders photorealistic visuals from real sensor data while supporting physics simulation through MuJoCo integration. It introduced the 'Gibson Database' of 3D reconstructions that became the standard for embodied navigation research. The environment supports multiple tasks including point-goal navigation, exploration, and object interaction, with a focus on bridging the gap between simulation and real-world deployment. Gibson laid the groundwork for subsequent platforms like Habitat and has been cited in hundreds of papers.