In a new paper published jointly by researchers at the University of California and Nvidia, the NaVILA Model is the core innovation of NaVILA that does not need a map in advance, the robot only needs to "understand" human natural language instructions, combined with real-time visual images and lidar information, and can autonomously navigate to the designated location. NaVILA not only gets rid of the dependence on maps, but also further expands the navigation technology from wheeled to legged robots, enabling robots to overcome obstacles and adaptive path planning in more complex scenarios.