DreamZero

The talk examines why robotics remains difficult, pointing to gaps in physical modeling, action representation, sensing, and hardware reliability. Researchers present approaches for longer tasks and adaptation using memory, embodiment-specific reasoning, and simulation-trained dexterous policies that transfer to real tools. Rerun co-founder Nico argues that robotics startups should solve a specific customer problem end to end, starting with teleoperation and off-the-shelf hardware, then iterating through real-world data and evaluation. General Intuition presents ways to make world action models faster enough for real-time robotics.

The talk explains sample efficiency and presents world models as a way for AI systems to learn and plan with fewer real-world examples. It contrasts differentiable control, where known physics enables optimization, with stochastic settings such as games, driving, and robotics, where other agents and large action spaces make reinforcement learning and test-time planning costly. It surveys model-based RL, synthetic rollouts, video-based world models, and JEPA’s prediction of future states in latent space. The speakers also discuss open problems including inaccurate physics, out-of-distribution adaptation, slow planning, and the scarcity of action and tactile data for robots.