From day one, mimic has been focused on a single goal: general-purpose dexterous manipulation. Today we're proud to announce the mimic hand M1 and the mimic wearable U1. We believe the only way to solve dexterous manipulation at scale is by going full-stack at the frontier of physical AI, building every layer ourselves around one fixed point, the human hand. The M1 is a highly backdrivable, tendon-driven hand that covers the full range of human capability, from heavy payloads to fine manipulation.

A technical dive inside our new "Midjourney Scanner"

7/ World models let systems imagine. These models internalize the physics and dynamics of reality (‘the true distribution’) so completely that they can predict future states and actions without ever touching the real world. This approach is the least tractable with today’s technology but also the most ambitious. They can enable models to learn across endless environments. Bottlenecks: 1) Realism - how realistic are the environments, physics, and actions that robots take? 2) Consistency - do objects in the scene persist over long periods of time? 3) Data - world models require the kinds of data that only massive players like Google and Meta have access to. World models are learned action models which can paint walls, move around, and even skydive! When massive video pre-training is layered with just dozens of hours of tele-op, a robot can achieve unprecedented generalization (Genie3, GAIA-2, VJEPA2). There’s a ton of interesting folks building here outside of robotics (like WorldLabs and General Intuition) and in the future, we expect them to be useful for robot learning.

A comprehensive, hands-on tutorial on the most recent advancements in robotics 🤟 ...with self-contained explanations of modern techniques for end-to-end robot learning & ready-to-use code examples using @LeRobotHF and @huggingface. Now available everywhere! 🤗