V-IRL: Grounding Virtual Intelligence in Real Life

An open-source framework for embodied agent and open-world computer vision research. Develop practical agents and test foundation models grounded with real street view imagery from around the world.