🥰Our paper SIMS: Simulating Stylized Human-Scene Interactions with Retrieval-Augmented Script Generation, has been accepted to #ICCV2025 📷 Our framework can generate stylized, emotional, physically plausible, and long-term human-scene interaction motions.

🔥Introducing SIMS: a new framework for simulating stylized human-scene interactions. 💺🚶 Congrats to Wenjia Wang on SIMS being accepted to #ICCV2025! Given a story prompt, SIMS uses retrieval-augmented generation (RAG) to produce long-term scripts, then drives

Now in Nature! 🚀 Our method learns a controllable 3D model of any robot from vision, enabling single-camera closed-loop control at test time! This includes robots previously uncontrollable, soft, and bio-inspired, potentially lowering the barrier of entry to automation! Paper:

Five Golden Nuggets from this talk: 1. Pretrained finetuned vs single task policy: - Because of training on different tasks the pretrained policy has more recovery behaviours. - Somehow the visual-action mapping across different tasks and environments leads to this behaviour.

Creative people and creative humanoid products!

Damn I want this and want to make it legged😂. A special thing is that this little robot will be powered by open-source and the infinite community. I love this insane idea.

Bro sharing humanoid training memos. So funny. I will join u.

People ask why humanoids. Here is why 🤣

Excited to share Flow Matching Policy Gradients: expressive RL policies trained from rewards using flow matching. It’s an easy, drop-in replacement for Gaussian PPO on control tasks.

One of our clients LightRobotics recently used our MiniHi robot to perform a cartwheel, demonstrating its outstanding agility and motion capabilities. Welcome to explore collaboration opportunities. #Robotics  #Geek  #humanoid The Humanoid Hub RoboHub🤖 T.Yamazaki CyberRobo

Got this little, cute humanoid robot to do cartwheels. Shout out to my mentor and the team at Light Robotics (a startup in China) 🤸‍♂️

Excited to open-source GMR: General Motion Retargeting. Real-time human-to-humanoid retargeting on your laptop. Supports diverse motion formats & robots. Unlock whole-body humanoid teleoperation (e.g., TWIST). video with 🔊

Summary of GPT-OSS architectural innovations: 1. sliding window attention (ref: arxiv.org/abs/1901.02860) 2. mixture of experts (ref: arxiv.org/abs/2101.03961) 3. RoPE w/ Yarn (ref: arxiv.org/abs/2309.00071) 4. attention sinks (ref: streaming llm arxiv.org/abs/2309.17453)

wow awesome perception and waypoint planning

🚀 ASAP is now FULLY open-source! 🚀 ✅ Humanoid RL motion tracking & delta actions ✅ Motion retargeting to any humanoid ✅ ASAP Benchmark motions + pretrained policies ✅ Sim2Sim & Sim2Real ready — run ASAP in sim or on your G1 robot! 🔗 github.com/LeCAR-Lab/ASAP

Looking forward to BFMs for humanoid robots😍

Check out the first PARC release! It comes with a helper program I'm working on called MOTION FORGE. It also comes with a generator and tracker model trained for 3 PARC iterations, as well as datasets. More updates will be coming soon! github.com/mshoe/PARC

Want to achieve extreme performance in motion tracking—and go beyond it? Our preprint tech report is now online, with open-source code available!

A very sci-fi scene: A robot walking a dog on the streets of Shanghai. A glimpse of the future family. 😂

the most impressive part of this? there’s no big secret. no groundbreaking method. this is what high quality data and excellent engineering looks like. huge congrats to the team! most interesting parts of the paper below: