It was a joy bringing Jason’s signature spin-kick to life on the @UnitreeRobotics G1. We trained it in mjlab with the BeyondMimic recipe but had issues on hardware last night (the IMU gyro was saturating). One more sim-tuning pass and we nailed it today. With @qiayuanliao and

HITTER in mjlab @kevin_zakka

Sleigh the season with the most personal gift around. Get them a Cameo video!

New video is out, teaching a Unitree G1 humanoid to walk using reinforcement learning (PPO). First time I've got sim2real to work with robotics, sharing what I've learned and testing out how good the policy actually is by walking around outside on some semi challenging terrain.

mink 1.0.0 is out! You can now enforce tasks as strict QP equalities. One neat application is "locking" certain DoFs. In this example, joints 1 & 2 of the Panda are locked when the cube is red and then unlocked when it turns green.

We use ty in mjlab ⚡️

“It’s really hard to both identify unique people and to do that in a privacy-preserving way…having a human-centric system turned out to be way more important in the world today than we realized.” - Sam Altman

Introducing FoundationMotion. A large-scale, video-derived motion annotation dataset & auto-labeling pipeline + advanced models for motion understanding. Fully open-source: code, datasets, and models, free to use and build on. Understanding motion is core to physical reasoning,

Action chunking is drawing growing interest in RL, yet its theoretical properties are still understudied. We are excited to share some insights on when we should use action chunking in Q-learning + a new algo (DQC) to tackle hard long-horizon tasks! https://t.co/izVWQBgH3c🧵1/N

Got the manipulation task in #mjlab working with the #SO101 arm. Only required a few modifications to @kevin_zakka's very elegant code. Trains in ~1-2 hours on a 3090 at ~70-100k steps/second. Looking forward to the pixel-in update and sim2real transfer. Code:

tyro 1.0 is out 🐣 This has been a pet project/niche interest of mine for ~4 years now, so it's a bit of a sentimental moment... https://t.co/bAibP3RjxE

We built a high-fidelity, cable-driven robotic hand in MuJoCo — accurately reproducing the full internal tendon system 🤖🧩 It’s now officially part of the MuJoCo Menagerie. On top of it, we developed a zero-shot RL training & real-world deployment pipeline based on MuJoCo

God sends a word from the future into the present, and everything changes. Watch Robin Bullock Follow for more

Introducing Correspondence-Oriented Imitation Learning (COIL), a self-supervised framework for robotic manipulation using 3D keypoint trajectories as a flexible, unified task representation!

We've setup nightly benchmarking for mjlab to track env/physics throughput and policy health over time. Every night we train + evaluate a tracking policy on the G1, measuring task performance and sim throughput across multiple tasks. https://t.co/gfLK3HBOUU

Shoutout to @mustafa_hdb who has been doing all the tiled rendering magic in mujoco_warp!

Edit: the policy above is trained with depth. Here's the RGB policy!

Big Tech Is Quietly Replacing Human Speech With AI Control Interview on The Vic Porcelli Show on @NewstalkSTL with @yovic7 & @kenjwilliams https://t.co/O7Bs6nT8SJ

Coming soon to mjlab and a long time in the making: RGB-D camera rendering! We can solve cube lifting with the YAM arm from 32×32 RGB frames in about <5 minutes of wall-clock time. Here's a clip showing emergent "search" behavior along with our upcoming viser visualization.

@kevin_zakka thanks, it's been a pleasure workin with mjlab!

Generative models (diffusion/flow) are taking over robotics 🤖. But do we really need to model the full action distribution to control a robot? We suspected the success of Generative Control Policies (GCPs) might be "Much Ado About Noising." We rigorously tested the myths. 🧵👇

Geoff the G1 preparing to go offroading IRL. I did a terrible job at the reward function here and was actually just tuning in to see all of what was broken and instead found a pretty good model. The robots just want to learn.

Happy to announce our neurips’25 paper, real world RL of active perception behaviors! I am pretty excited about this project - I learned that real world robot RL is actually quite straightforward. Details below:

For any PhD students looking for a summer internship to work with me on RL, VLAs and humanoid robots in SF, please consider applying. https://t.co/adODiI9gja

Check it out here: