AI now enables robots to adapt rapidly to changing real world conditions @AIatMeta

Uploaded July 2021 | Updated September 2026, 14 hours ago

Robots can be programmed to do impressive things. But for the first time, with AI, robots can learn from scratch how to walk in different environments. Researchers from Facebook AI, Berkeley AI Research, and Carnegie Mellon University’s School of Computer Science have developed Rapid Motor Adaptation (RMA), a breakthrough in artificial intelligence that enables legged robots to adapt intelligently and in real time to challenging, unfamiliar new terrain and circumstances, much like humans do.

Video courtesy of Facebook AI Research. Robot test footage provided by researchers from Berkeley AI Research, Facebook AI Research and Carnegie Mellon University.

AI now enables robots to adapt rapidly to changing real world conditions

New Research: Molecular Property Prediction, Language Processing, and Neuroscience | AI at Meta

How can we teach AI skills using visual demos? Model-based Inverse Reinforcement Learning

Advancing the state of the art in computer vision with self-supervised Vision Transformers

Habitat 2.0: Training home assistants to rearrange their habitat

Yann LeCun on the future of deep learning hardware

Is human led mathematics over? Panel with Joelle Pineau, Timothy Gowers & Yann LeCun | Meta AI

Reimagining Metas Infrastructure for the AI Age | AI at Meta

Introducing SAM 3D: a New Standard for 3D Object & Human Reconstruction from a Single Image