Explore RobotsBeat's latest reporting, research, and expert analysis tagged with "video understanding", collected in one continuously updated archive.
1X has introduced a world model that enables its NEO humanoid robot to learn new tasks by observing videos. The approach aims to reduce reliance on manual programming and task-specific data collection.
The Allen Institute for AI has released Molmo 2, an open multimodal model designed to deliver stronger video and image understanding with significantly less training data. The model targets robotics, automation, and real-world perception tasks.