
Ph.D. student at UPenn @GRASPlab | Prev Master student @UCSDJacobs, Undergrad @sjtu1896 | Working on Embodied AI & Robot Learning.
Philadelphia, PA
Joined May 2022
- Tweets129
- Following491
- Followers439
- Likes364
Pinned Tweet
Can we build a robot policy foundation model🤖 that incorporates 3D observations?
🚀Introducing FP3, a first 3D foundation policy for robotic manipulation. Using only 80 demonstrations, FP3 can solve a new task with over 90% success rates in the wild!
Let's dive in! 🧵1/N
From one hand to one thousand hands, how far can we further push it?
1000 hands, cross embodiment grasping .
🙌 XDex is at #CoRL2026 today!
Poster 117@ JW Marriott, Griffin 5:00–6:30 PM
keyiw.github.io/xdex/
Geng Chen retweeted
RoboSkills agent-as-policy is now in the real world! 🤖
- Hard coordination: twisting the lid off a bottle.
- Precise action: picking a slim pen and stacking, with no crashes.
- Cross embodiment: one workflow for both grippers and hands!
code: github.com/openrobokit/robos…
From tabletop to whole-body grasping!
Find it. Reach it. Grasp it. Easy for us, hard for a humanoid.
Introducing HuG (#CoRL2026): Learning Humanoid Whole-Body Affordance-Aware Grasping with Active Perception.
The robot looks around for a half-hidden object, squats or stretches its whole body to reach it, and grasps. Zero real demos, zero-shot to unseen objects and everyday scenes, from low tables to high shelves.
🌐 xuanbinpeng.com/hug/
5
215
Really excited to see this finally released! So much useful stuff in OpenRoboKit, from IK and retargeting to optimization and grasping. Well worth trying out for your next robotics project!
Excited to share OpenRoboKit, our series of open-source robotics projects! 🤖
- RoboKit for fast, scalable optimization & simulation github.com/openrobokit/robok…
Our RoboKit paper has been accepted for a spotlight at #CORL2026
- RetargetingKit for motion transfer & teleoperation github.com/openrobokit/retar…
- GraspKit for end-to-end grasping github.com/openrobokit/grasp…
Website: openrobokit.com
Check the video (sound on 🎵) for a 16-hour grasping time-lapse in beautiful San Diego, plus a quick look at dexterous retargeting and robot simulation.
12
490
It turns out that with a minimal set of tools and workflows, agents can generate super cool motion for animation characters 💃 and humanoid robots 🤖, even with hand-object interaction!
Given text, images, or videos, the agent selects motion models, builds the scene and object, retargets to the character or robot, and refines poses, timing, and contacts from rendered feedback.
Try our motion generation skill in RoboSkills! github.com/openrobokit/robos… We also provide a fixed agent-free workflow for motion without interaction.
Also check out other great skills for robotics in RoboSkills openrobokit.com/roboskills including Real2Sim, Auto Research, Agent as Policy, and Agentic Simulation!
teapawt.net/jianglong_ye/status/21…
We introduce RoboSkills: a minimal, low-cost skillset for agentic robotics 🤖
It supports diverse robotics workflows, such as learning cat’s cradle 🧶 from a human video (Real2Sim) and transferring multi-person motion to characters and humanoids 💃 (Motion Generation).
We also support Auto Research, Agent as Policy, and Agentic Simulation. With suitable tools and workflows, we enable cheaper models to solve complex tasks.
GitHub: github.com/openrobokit/robos…
Website: openrobokit.com/roboskills
Watch the video for more examples.
3
316
Human manipulation has never been just about the hands. It's the whole body. Really impressive work and demo!
Humanoids today mostly manipulate objects with their hands. As humans, we use much more than our hands: we brace boxes against our chests, tuck objects under our arms, and roll balls along our forearms.
Introducing Whole-Body Dexterity (WBD), turning the humanoid body into one large dexterous manipulator.
wholebodydex.com 🧵👇
Geng Chen retweeted
Our G1 robot found its way to the pantry, picked up a Pringles can, and handed it over. No pre-built map. No training for this task.
Behind it is a coding agent, writing, running, and fixing the robot's programs as it goes. And it self-evolves: what works is saved as code for the next task, no retraining needed. This is ASENA.
Joint work with @Isabella__Liu, @edmund_bu, Johan Bjorck, @yin_hongxu, @zhengyiluo, @jankautz, @DrJimFan, @yukez, and @Sifei30488L. 🧵👇
Geng Chen retweeted
Can we scale robot learning without scaling human labor? 📈
Demonstrations, resets, failure recovery, constant supervision—collecting robot data still takes substantial human effort.
🔑 Three keys let agents take on more of the work:
1. Real2Sim2Real: Agents explore and learn in digital twins, where trial and error is safer and cheaper.
2. Autonomous Recovery: Robots learn to recover from unexpected failures instead of waiting for human resets.
3. Human-on-Demand Learning: Multiple robots run in parallel, and humans step in only when autonomous recovery falls short.
🥑 Recova brings these together to get more robot learning out of every hour of human effort.
research.nvidia.com/labs/gea…
Joint work with @anjjei, Johan Bjorck, @ZhidingYu, @yin_hongxu, @jankautz, @DrJimFan, @yukez, and @Sifei30488L.
👇 Here's how it works 🧵
Congrats on the new journey!
Three months ago, I wouldn’t have imagined we could build all of this so quickly. Seeing the infrastructure, research agents, and robot learning framework come together—and start producing real results—has been incredibly rewarding.
What excites me most is where this could lead. As agents take on more of the research loop, we can explore more ideas, learn from more failures, and spend more time asking the questions that matter. Each experiment can leave us with better models and a better system for doing the next one.
Proud of what this team has built. There’s still a lot to figure out, but these three months have made me much more optimistic about what we can build next.
1
253
Still remember chatting with @YutongLiang_ on his new project when he brought up this idea of a modern revisit of the classic OpenAI Rubik’s Cube challenge.
Super cool to see him turning that idea into reality within just a few months!
Meet FINGR: a dexterous hand that solves a 2×2 Rubik’s Cube with continuous finger tricks.
Here’s how we got the fingers to work together. 🧵
1
6
492
Geng Chen retweeted
When we started HARMONY six months ago, we were curious how far agentic approaches could take 3D reconstruction. The field has advanced quickly, and we see strong potential in combining high-level reasoning with geometric optimization. We’re releasing an open-source framework competitive with Astra in reconstruction quality, together with the HARMONY300 benchmark. Proud of @sofine_sun and @chenwangcw!
Excited to share our new 𝗼𝗽𝗲𝗻-𝘀𝗼𝘂𝗿𝗰𝗲 work, 𝗛𝗔𝗥𝗠𝗢𝗡𝗬, a framework for reconstructing compositional 3D scenes from single indoor images.
Much like a 3D artist, our VLM agent builds the scene in a 𝗵𝗶𝗲𝗿𝗮𝗿𝗰𝗵𝗶𝗰𝗮𝗹 𝗼𝗿𝗱𝗲𝗿: first the room structure and camera, then wall-mounted objects, free-standing furniture, and finally dependent decorations. After each stage, it compares its render against the input image, reflects on what went wrong, and iteratively corrects the scene to prevent errors from accumulating.
Even with older VLM backbones, our framework remains 𝗰𝗼𝗺𝗽𝗲𝘁𝗶𝘁𝗶𝘃𝗲 𝘄𝗶𝘁𝗵 𝗚𝗣𝗧-𝟲 𝗔𝘀𝘁𝗿𝗮 𝗶𝗻 𝘃𝗶𝘀𝘂𝗮𝗹 𝗾𝘂𝗮𝗹𝗶𝘁𝘆, 𝘄𝗵𝗶𝗹𝗲 𝗮𝗰𝗵𝗶𝗲𝘃𝗶𝗻𝗴 𝗯𝗲𝘁𝘁𝗲𝗿 𝗴𝗲𝗼𝗺𝗲𝘁𝗿𝗶𝗰 𝗮𝗹𝗶𝗴𝗻𝗺𝗲𝗻𝘁 for single-image 3D scene reconstruction.
We also release 𝗛𝗔𝗥𝗠𝗢𝗡𝗬𝟯𝟬𝟬, a benchmark for single-image 3D scene reconstruction with 300 scenes across three difficulty tiers: easy, medium, and complicated.
🔗 Project: cwchenwang.github.io/harmony
📊 Leaderboard: cwchenwang.github.io/harmony…
🤗 Data: huggingface.co/datasets/Shuf…
📄 Paper: arxiv.org/abs/2609.26793
Geng Chen retweeted
Children learn to walk by falling. Robots fail too—but we usually throw those failures away.
Yet failed rollouts tell us something successful demonstrations cannot: what bad actions actually lead to.
So, can robots learn from their failures too?
Introducing FACT: Failure-Aware Causal Training for World-Action Models.
🔗website: fact-wam.github.io/
📄arxiv: arxiv.org/abs/2608.10232
💻 code: github.com/Bariona/FACT
Geng Chen retweeted
Human demos contain rich hand-object skills. The challenge is the embodiment gap. Human and robot hands differ in shape, joints, and contacts.
Introducing ConTrack, our latest work that turns contact-rich human hand-object demos into robot motions.
Excited to announce that FP3 has been selected as an ICRA 2026 Best Paper Award Finalist on Robot Learning! 🎉
Unfortunately, I won’t be able to attend in person in Vienna, but our advisor @gao_young will be presenting the work today. Please stop by if you are at ICRA!
📍 Session: TuAT1.3
🕦 Time: June 2 11:20–11:30
Glad to share that our paper FP3 has been accepted to #ICRA2026! 🎉Huge thanks to my co-authors and collaborators, it's been a long ride.
In FP3, we demonstrate that 3D perception is a strong ingredient for learning robot foundation models. It’s been about 1 year since we wrapped up this work, and during this year, video world models have advanced insanely fast to push the vision community into a “do we still need explicit 3D?” debate. The same question also applies to robotics. Classic stacks rely on rigorous geometry, but will end-to-end learning make implicit 3D sufficient?
Excited to explore the real boundary.
1
15
1,693
A hierarchical policy is essential for long-horizon tasks!
VLA/VAs are doing well on short skills like pick-and-place. But real tasks rarely stop after one action, they require 1) many interdependent steps, 2) progress tracking, and 3) recovery from mistakes.
In our paper LoHo-Manip, we address long-horizon manipulation with trace-conditioned VLA planning: a task manager tracks what’s done, plans what remains, and guides execution with visual traces.
Strongly bullish on scaling human data for robotics! Feel excited to help a bit in this ambitious project.
Introducing EgoVerse: an ecosystem for robot learning from egocentric human data.
Built and tested by 4 research labs + 3 industry partners, EgoVerse enables both science and scaling
1300+ hrs, 240 scenes, 2000+ tasks, and growing
Dataset design, findings, and ecosystem 🧵
2
3
817















