I will join UChicago CS @UChicagoCS as an Assistant Professor in late 2026, and I’m recruiting PhD students in this cycle (2025 - 2026). My research focuses on AI & Robotics - including dexterous manipulation, humanoids, tactile sensing, learning from human videos, robot

Huge thanks to my brilliant collaborators: Jessica Yin, @WiYoungsun, Sayantan Kundu, Mike Lambeta, William Yang, @ChanghaoWang78, @TingfanW1208, @JitendraMalikCV, and @t_hellebrekers! Check out the website ( https://t.co/gh76bWyTha) for the full details, code, and design files!

In our real-world wiping task, our tactile-aware policy achieved a 72% success rate, significantly outperforming vision-only baselines. The vision-only policies frequently failed due to contact-related issues (e.g., inconsistent pressure or lost grip on the sponge) that tactile

The most exciting part: We trained a robot policy for a wiping task exclusively on human demonstrations collected with OSMO - without any real robot data! By using the same OSMO glove on both the human and the Psyonic Ability Hand, we minimize the visual and tactile "embodiment

We designed OSMO to be "in-the-wild" compatible: - It integrates with state-of-the-art hand tracking methods, including wearable AR/VR headsets (Quest 3, Aria Gen 2, Apple Vision Pro) and generic RGB videos. - The glove's appearance (low-profile electronics, beige fabric) is

OSMO's design is all about bridging the gap between human and robot dexterity! It has 12 three-axis magnetic tactile sensors across the fingertips and palm, measuring both shear and normal forces (what vision can't see!). It preserves full human dexterity, allowing for natural,

Human videos are great for robot learning, but they critically lack the sense of touch needed for complex manipulation tasks! Introducing OSMO, an open-source tactile glove that captures rich contact signals during human demonstrations for direct, successful transfer to robots.

Thank you, @YiMaTweets! It’s been an honor to be your student! I wouldn’t be here without your guidance.

Simulation has been a central theme of my PhD, but one thing has always troubled me (and many colleagues): complex physics and rich sensing are incredibly hard to simulate. In this work, we explore a different path: learn a simplified task in simulation, then use that

I'm recruiting multiple PhD students this cycle to join me at Harvard University and the Kempner Institute! My interests span vision and intelligence, including 3D/4D, active perception, memory, representation learning, and anything you're excited to explore! Deadline: Dec 15th.

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. 🧵👇

Can we quickly improve a pre-trained robot policy by learning from real world human corrections? Introducing Compliant Residual DAgger (CR-DAgger), a system that improves policies performance to close to 100% on challenging contact-rich manipulation problems, using as few as

Thank you @Suddhus! It was a great honor working with you!

Sim-to-real learning for humanoid robots is a full-stack problem. Today, Amazon FAR is releasing a full-stack solution: Holosoma. To accelerate research, we are open-sourcing a complete codebase covering multiple simulation backends, training, retargeting, and real-world

Thank you @ken0324! I'm also super excited to join!

Thank you @drjingxi!

Thank you @plopesresearch! Looking forward to working together!

Thank you @irmakkguzey! It was a truly great and productive summer! Learned a lot from you.

Thank you @mangahomanga! It’s an honor to work with you!

Thank you Jianren for the kind words, really appreciate this!