Sizhe Wei

Robotics Ph.D. student @GeorgiaTech

[Sizhe Wei @SizheWei]

Looking forward to present MS-HGNN at #L4DC next week in Ann Arbor. Come to say hi if you are around 🤖😁

[Lunar Lab @lunarlab_gatech]

It's Friday afternoon, and we’re packing for L4DC 2025 @l4dc_conf next week 😎

Excited to share MS-HGNN — our take on bringing morphological symmetry into GNNs for legged robot dynamics.

🧵 (1/4) We bake both kinematics and morphological symmetry into the model →
- Generalizable
- Sample-efficient
- Compact

🌐 Project: https://t.co/1qIqrl0ziX
📄 arXiv: https://t.co/PEgBaIQKj1
💻 Code: https://t.co/Ykb7yXQK1H

Team: @xie_fengze32905, @SizheWei, Yue Song, @yisongyue, @ganlumomo

See you next week in Ann Arbor! #Robotics #L4DC2025 #GeorgiaTech

[Fei-Fei Li @drfeifei]

https://t.co/l3mSj9Ur24

[Maximilian Du @du_maximilian]

When a robot learns a new behavior, what keeps it from forgetting the ones it already knows?

Prior work, especially on VLAs, shows that rehearsing past experiences can enable strong continual learning. But why does rehearsal work so well, and when does it fail?

We find that a small but ✨special subset✨ of rehearsed data largely determines whether old behaviors are retained or forgotten. We call these examples Memory Anchors. ⚓

Withholding just the top 10% of Memory Anchors from rehearsal increases forgetting by up to 4.5x. Increasing their presence, meanwhile, reduces task forgetting and enables a real robot to learn challenging task sequences.

Website: https://t.co/JhGATqmQ6F

Paper: https://t.co/i3PPO2NxP3

Curious? Read on! 🧵👇 (1/9)

[Yu Xiang @YuXiang_IRVL]

100K Hours = 4,166.67 days = 11.4 years real-world trajectories

57% success rate on the RoboCasa365 benchmark

Maybe we're missing something

[Jensen Huang @JensenHuang]

For my first post, I’m sharing a letter @NVIDIA signed on why open models matter.

AI will transform every industry, power every company, and be built by every country.

Open models strengthen safety and cybersecurity, accelerate innovation and diffusion, and enable sovereignty.

The world needs both frontier closed models and frontier open models.

https://t.co/AUKzoQ5Ikb

[Maximilian Du @du_maximilian]

Would you want a robot dog as a personal trainer? We built Snoopie 🐶, a quadruped robot that runs interval workouts with you by providing embodied feedback to pace you 🏃
🔗 https://t.co/9ICqy7RhQQ
Here’s what we found when we took it to the track… 🧵1/7

[Frank Dellaert @fdellaert]

https://t.co/LOl8mxL5Lm

[Pete Florence @peteflorence]

https://t.co/cwqP98pKCq

[Generalist @GeneralistAI]

Introducing GEN-1.

Our latest milestone in scaling robot learning.

We believe it to be the first general-purpose AI model to master simple physical tasks.

99% success rates, 3x faster speeds, adapts in real time to unexpected scenarios, w/ only 1 hour of robot data.

More🧵👇

[Sizhe Wei @SizheWei]

🎉🎉🎉

[Lunar Lab @lunarlab_gatech]

Excited to share that we have 3 papers accepted to ICRA 2026, 1 paper accepted to the CVPR 2026 Findings Track, and 1 paper presented at 3DV 2026! 🎉 🎉

[1] GaussianFormer3D: Multi-Modal Gaussian-based Semantic Occupancy Prediction with 3D Deformable Attention - https://t.co/EhB9eU38NM
Lingjun Zhao, Sizhe Wei @SizheWei, James Hays @jhhays , Lu Gan
ICRA 2026, RSS 2025 Gaussian Representations for Robot Autonomy Workshop

[2] STATE-NAV: Stability-Aware Traversability Estimation for Bipedal Navigation on Rough Terrain - https://t.co/iNZbXfgW4K
Ziwon Yoon, Lawrence Y Zhu @LawrenceZhu22, Jingxi Lu, Lu Gan, Ye Zhao
IEEE Robotics and Automation Letters 2025, ICRA 2026

[3] A Generalizable Physics-guided Causal Model for Trajectory Prediction in Autonomous Driving - https://t.co/wz7yDLE9VR
Zhenyu Zong, Yuchen Wang, Haohong Lin, Lu Gan @ganlumomo, Huajie Shao @HuajieShaoML
ICRA, 2026

[4] ShelfGaussian: Shelf-Supervised Open-Vocabulary Gaussian-based 3D Scene Understanding - https://t.co/hx0gQ1MEd7
Lingjun Zhao*, Yandong Luo*, James Hays @jhhays, Lu Gan
CVPR 2026 Findings

[5] Diffusion-Denoised Hyperspectral Gaussian Splatting - https://t.co/gJ0K6Gm1N7
Sunil Kumar Narayanan @gaussiannerf, Lingjun Zhao, Lu Gan @ganlumomo, Yongsheng Chen
3DV, 2026

Huge congratulations to all authors and collaborators. Looking forward to connecting in Vienna and Denver!

#lunarlab #Robotics #ICRA2026 #CVPR2026 #3DV2026 #AutonomousDriving #humanoid

[Yanjie Ze @ZeYanjie]

Agree &
Novelty is not needed for robotics. Engineering + good taste is what you need.

[RAI Institute @rai_inst]

Atlas can perform a backflip combo like a gymnast and also walk naturally across a stage.

These very different movements are enabled by the whole-body learning framework developed by the RAI Institute and deployed by @BostonDynamics. We’re getting closer to robust, generalist humanoid behaviors that transfer zero-shot from simulation to physical performance: https://t.co/vqM7VxNtds

[Zhengyi “Zen” Luo @zhengyiluo]

Wow thoroughly impressed by this demo.

First time seeing a humanoid company mentioning > 1,000 hours of human motion data as one of the core thing that made this happen.

Scaling up motion tracking works!

[Chen Tessler @ChenTessler]

MaskedMimic policy for the G1 is now live in https://t.co/cwjbPglhON

Give it a try and share your experiences with us so we can continue improving!

Docs: https://t.co/RASwWtyfLb

[Brett Adcock @adcock_brett]

Today we're introducing Helix 02

Dancing robots are trivial, the hard part is intelligent control

This is our most powerful model to date - able to work across complex tasks & long time horizons
https://t.co/cExYWUUoDp

[Chenhao Li @breadli428]

🌎World models can predict, but controlling real robots from imagination sees a long-standing failure due to hallucination.

🧠Introducing Uncertainty-Aware RWM: a black-box, end-to-end neural dynamics model with long-horizon uncertainty propagation.

🎯https://t.co/CQgvuEWTaQ

[Moo Jin Kim @moo_jin_kim]

We release Cosmos Policy 💫: a state-of-the-art robot policy built on a video diffusion model backbone.

- policy + world model + value function — in 1 model
- no architectural changes to the base video model
- SOTA in LIBERO (98.5%), RoboCasa (67.1%), & ALOHA tasks (93.6%)

🧵👇

[Eric Jang @ericjang11]

Life update: I've decided to leave 1X.

It's been an honor helping grow the company. I joined Halodi Robotics in 2022 (prior name of the company) as the only California-based employee. At the time, we were about 40 based out of Norway and 2 in Texas. My first hire and I worked from my garage for a few months to save money. Today, 1X is hundreds of people, with hardware, design, software, AI, manufacturing, product all relocated to the SF Bay area, firing on all cylinders and working on getting NEO ready for the home. A big thank you to all my colleagues that I worked with.

It was a hard decision to leave. When working at an exciting startup that is growing fast, there's always so much to do and never a perfect time time to move on. We have several works in the pipeline that are so exciting because they greatly advance general autonomy and scalability of our deployment approach and really show a realistic path towards the product working. The recent World Model autonomy update is one example, and there's more coming. The 1X factory is so exciting. Things are accelerating at a speed I would have been surprised by a few years ago.

In 2022, most technologists and researchers and VCs were skeptical about humanoids and large scale imitation learning. "Why Legs?" "How could end-to-end learning ever be good enough?" "Why go for the home and not the factory?" "How will we ever gather enough data?"

The Overton window on general-purpose robotics has shifted a lot since then. Although we are still early in our mission, I remain confident that soon, house robots will be as commonplace as air conditioners, cars, and ChatGPT. Just talk to the bot, and it will go and quietly get it done. Entire economies will eventually re-organize around this technology. People get it now.

What's next?

I believe that progress in applied deep learning generally rides on "harnessing the magic" of a few magical objects. These magical objects possess way more generalization power than one might normally expect. Just asking the LLM to understand what you want is magic. Video generation models are magic. Reasoning is magic. You don't run into a magic object every day, but when you do, you make sure to grab it and put it to work to make something useful in the robot somehow.

A lot of my early conviction for where robotics was headed was working on BC-Z from 2018-2021. The "magical object" I bet on at the time was the surprising data-absorption capabilities of supervised learning and "just ask for generalization". This pioneered a lot of the standard ingredients we see in VLAs today:
- Generalization to unseen language commands
- Human-Guided DAgger for policy improvement
- Open-loop auxiliary predictions + receding horizon control, AKA action chunking
- Manipulation keypoints to improve servoing
- Simple ResNet18 with FiLM conditioning on multi-modal inputs

The next "magical object" we bet on at 1X was video models, because they are clearly magical objects that learn a data distribution not too dissimilar from what a robot needs to learn. They generalize surprisingly well.

I am once again feeling that there are more magical objects in play now, which opens up a lot of new possibilities for robotics and beyond. I'm taking a few months to empty my cup of priors and gain fresh perspective. When I left Google in 2022, I spent about 2 weeks deciding what to do next. This time, I want to take a lot more time to catch up what has happened in the broader AI + robotics space.

I've been re-implementing some deep learning papers. I'm working on a big tutorial for my blog. I'm learning all the Claude power user tricks. I'm reading the Thinking Machines blog posts to understand what kinds of experiments are being run at frontier labs. I'm reading Ben Katz's 2016 thesis on the Mini-cheetah actuator. I'm traveling to China in March to meet incredible companies in the Chinese robotics ecosystem. Now, more than ever, is the time for both humans and machines to learn. The next token of my life sequence will be an important one.

To colleagues and investors that bet on 1X early, even before we became a household name - I thank you from the bottom of my heart. I won't forget it♥️

[Chen Geng @gengchen01]

Generative AI has seen remarkable success in generating static 3D shapes. Yet, building a comprehensive 4D World Model requires a crucial missing piece: realistic dynamics.

CHORD presents a universal framework to bridge this gap:

Input: Static 3D objects (zero dynamic annotations). Output: Plausible 4D object motion & coherent 4D scenes.

No rigs or skeletons required. 🚫🦴

🧵: 2/n

[Tae Hoon Yang @taeyang___11]

Excited to introduce Locomotion Beyond Feet!! A whole-body locomotion system that enables humanoid robots to crawl, climb, and recover using hands, knees, elbows, and torso, extending locomotion beyond legs alone.

Project Page: https://t.co/catSmExed6

1/7

[Wenlong Huang @wenlong_huang]

What if we can simulate an *interactive 3D world*, from a single image, in the wild, in real time?

Introducing PointWorld-1B: a large pre-trained 3D world model that predicts env dynamics given RGB-D capture and robot actions.

🌐 https://t.co/ShGZm3hAWi

from @Stanford @nvidia