Robotics is entering a new phase. VLA models, embodied reasoning, world models, teleoperation-scale data collection, simulation, RL, and humanoid platforms are all advancing at once. The question is no longer whether foundation models will matter for robotics; it is whether robotics is approaching its own GPT moment.
The analogy, however, is imperfect. Unlike language, the physical world is expensive to sample, hard to evaluate, and unforgiving when models fail. Robots need more than semantic understanding: they need 3D perception, physical causality, force awareness, proprioception,…

Hosted by Jeson Lee, Akshay Bhushan & Sophia Lu