The Toolmakers Build the Body

The creators of LeRobot - the open-source toolkit powering robotics worldwide - just unveiled Northstar, a humanoid robot that learns from human demonstration instead of manual programming. Rémi Cadène's team at UMA includes Tesla Optimus engineers, DeepMind researchers, and the founders who gave the entire industry its software foundation. Now they're applying that proven infrastructure to build the body. This is systems-level thinking that inverts how humanoid robots get built - starting from learning architecture, not hardware speculation. In this episode: - UMA Northstar humanoid robot: Real-Time Learning architecture explained - Why starting from software infrastructure changes everything about hardware - The founding team's pedigree: Tesla, DeepMind, and LeRobot creators - How robots learn new skills from human demonstration - 40 million in seed funding from Greycroft, Relentless, and Unity Growth - Europe-first strategy targeting logistics, manufacturing, and healthcare - Why infrastructure builders making their own products represents a new competitive threat - Test prototype planned for late 2026 - Advisers including Yann LeCun (Meta) and Thomas Wolf (Hugging Face) The hard problem in humanoid robotics isn't actuators or joints - it's learning architecture and skill acquisition. UMA's team didn't just observe this problem. They solved it for everyone else first. Now they're applying those solutions to their own machine. Subscribe to Tomorrow Unveiled for insights into technologies shaping our future. #UMANorthstar #HumanoidRobot #RealTimeLearning #RoboticsAI #LeRobot #PhysicalAI #RobotLearning #DeepMindRobotics #OpenSourceRobotics #TechInnovation Watch on YouTube: https://www.youtube.com/watch?v=NwAP2PtOaMM