🔥Telegram İfşa

I like the idea of shifting human effort from data collection to creating realistic simulations. There are many routes to scaling!

Reward models that help real robots learn new tasks—no new demos needed! ReWiND uses language-guided rewards to train bimanual arms on OOD tasks in 1 hour! Offline-to-online, lang-conditioned, visual RL on action-chunked transformers. 🧵

Super cool work by Jiaheng on performing RL efficiently in the real world!

Introducing VGC-Bench, a new plug-and-play benchmark made to support research efforts in the domain of Pokémon VGC! Read on for the open source code, the paper, and everything you need to dive in. [1/9]

Meet Casper👻, a friendly robot sidekick who shadows your day, decodes your intents on the fly, and lends a hand while you stay in control! Instead of passively receiving commands, what if a robot actively sense what you need in the background, and step in when confident? (1/n)

Imagine a future where robots are part of our daily lives — How can end users teach robots new tasks by directly showing them, just like teaching another person? 🧵👇

To truly learn what innovations are essential for improving generalist robot policies, we need to run MANY unbiased evaluations in a variety of environments 🎯 Presenting RoboArena: by crowdsourcing double-blind robot evals, we make them fair and scalable!

We've extended the submission deadline to the CoRL Workshop on Generalist Policies in the Wild to September 17th! sites.google.com/view/corl-robo…

Checkout Rutav's latest work on in-context learning from human videos. A few human demos at test time can quickly adapt your policy to the given task!

Checkout our #CoRL work on mixing data subsets for efficient policy training on downstream target tasks. Sateesh will be at CoRL presenting this work, so say hi if you're around and wanna learn more!