Arm + hand example coming soon to mink!

Our robot first trains in a simulated environment, which can model the physics of table tennis matches accurately. Once deployed to the real world, it collects data on its performance against humans to refine its skills back in simulation - creating a continuous feedback loop.

Heading to the RL_Conference to present our work Semi-Supervised One Shot Imitation Learning 🏹 More details below! Looking forward to catching up and chatting with old friends and new, let me know if you’re around! #RLC2024

🚀🤗

Can we make Transformers better and more efficient for robot learning? Excited to introduce Body Transformer (BoT), an architecture that leverages robot embodiment in the attention mechanism, by treating it as a graph of sensors and actuators.

The power of open source 🚀 Nice job Omar Rayyan !

Amazing results, I'm super impressed!

Today was the last of my internship at Boston Dynamics. I’m deeply grateful to my host Lucas Manuelli who has been an amazing mentor and friend this summer. I got to learn and absorb from world-class scientists and engineers over the last three months and it’s energized me as I

Control physics-based characters at interactive rates to climb over tables and more. No learning! Key ideas: (a) partwise MPC: switch between full-body and factored body-part MPC as convenient; (b) strategy guidance via sparse contact keyframes. SCA 2024 cs.ubc.ca/~van/papers/20…

Really neat work that builds on RoboPianist, eliminating the need for fingering labels and creating a large dataset for studying bimanual dexterity!

A very well written and concise overview of actuator design principles by Philipp. I'm lucky I sit next to him and can ask him about these things for free 🙃

"We design and manufacture low-cost and easy-to-use devices for collecting human demonstration data" 🤔 they forgot to mention git clone umi-gripper.github.io

To be clear, dissemination + innovation are great but having an entire release statement pass it off as an original creation is not cool.

📢 Today's my first day at Cornell Tech :) I will continue my full-time position as a scientist at Meta, and am teaching ML here on the side for the semester with Volodymyr Kuleshov 🇺🇦. The course is open source and you can follow everything in this repo: github.com/kuleshov/corne…

When we created RoboPianist, our goal was a benchmark that could be easily extended without data limitations. This work, led by Cheng Qian, shows how we can tap into one of robotics’ richest data sources, YouTube, to train an impressive generalist piano-playing agent!

Left and right models of the LEAP hand are now available in Menagerie. Thanks to ⁦Kenny Shaw⁩ for all the help 😀 github.com/google-deepmin…