Pinned Tweet
With differentiable physics-based simulation&rendering, our robot develops a “model” and learns to select informative viewpoints to perform needle threading (a deformable version of peg insertion).
Excited to share SAM-RL, sensing-aware model-based reinforcement learning via differentiable simulation and rendering. It allows a robot to select viewpoints and accomplish the task of needle-threading.
w/Yunhai, Cheng, @shuangz, @linshaonju, @Cewu_Lu
Web sites.google.com/view/sam-rl
Lin Shao retweeted
Remember TidyBot?
We developed an easy-to-assemble, customizable, open-source and low-cost version of our holonomic mobile base.
If you are at #CoRL2024 come see our live demo on the 2nd floor and come to Poster 30 at Session 3.
xitter.cf/jimmyyhwu/status/16569…
When organizing a home, everyone has unique preferences for where things go. How can household robots learn your preferences from just a few examples?
Introducing 𝗧𝗶𝗱𝘆𝗕𝗼𝘁: Personalized Robot Assistance with Large Language Models
Project page: tidybot.cs.princeton.edu
Lin Shao retweeted
Want a robot that learns household tasks by watching you?
EquiBot is a ✨ generalizable and 🚰 data-efficient method for visuomotor policy learning, robust to changes in object shapes, lighting, and scene makeup, even from just 5 mins of human videos.
🧵↓
Lin Shao retweeted
3 papers in #CoRL_2024! Come and talk with me about world model, equivariant learning, and dexterous manipulation!
RiEMann: Poster Session 1
Workshops: LEAP, X-Embodiment, MAPoDeL, LFDM.
D(R,O)-Grasp gets Best Robotics Paper on MAPoDeL workshop! – at Garching bei München, Deutschland
Lin Shao retweeted
We are excited to share our #CORL2024 paper (oral) on "Learning Quadruped Locomotion Using Differentiable Simulation" done in collaboration with Sangbae Kim @MIT. We present a new way to learn to walk in minutes without parallelization, outperforming PPO in sample efficiency!
PDF: arxiv.org/pdf/2403.14864
Video: youtu.be/weNq_w715xM
We present a new framework for learning quadruped locomotion. By leveraging differentiable simulation for policy optimization, our approach achieves fast convergence and stable training, significantly outperforming model-free #ReinforcementLearning methods like PPO in sample efficiency. The key enabler is to combine a high-fidelity, non-differentiable simulator for forward dynamics with a simplified surrogate model for gradient backpropagation. Our framework enables learning quadruped walking in simulation in minutes without parallelization. When augmented with GPU parallelization, our approach allows the quadruped robot to master diverse locomotion skills on challenging terrains in minutes.
This work highlights one of the first successful real-world applications of differentiable simulation for quadruped robots, offering a compelling alternative to traditional RL methods.
Kudos to @realyunlong!
@UZH_Science @UZH_en @UZHspacehub @uzh_ifi @ERC_Research @MITMechE
Lin Shao retweeted
Check out our paper: arxiv.org/abs/2403.19460
Code: github.com/HeegerGao/RiEMann
Awesome cooperation with @ZhengrongX, Shuying Deng, Tianhai Liang, Siqi Yang, @linshaonju , and @HarryXu12.
Lin Shao retweeted
Folding clothes with $250 robot arms. I've added another motor to improve mobility and extend the reach. The CAD files and the code are public at: github.com/AlexanderKoch-Koc…
(video at 2x speed)
Lin Shao retweeted
"Why don't we have better robots yet?": just posted on the @TEDTalks home page under Newest Talks (3rd row from the top) with links to @PieRobotics and @Forbes article on art by Ben Wolff @creativecellist @Berkeley_AI @UCBerkeley @Cal_Engineer TED.com
Lin Shao retweeted
Current 3D generative models are slow and low quality. We present GRM, a large-scale model that reconstructs 3D Gaussians in 0.1s and generates high-quality 3D assets from text or single images in a few seconds.
justimyhxu.github.io/project…
Demo: huggingface.co/spaces/GRM-de…
1/4
Lin Shao retweeted
Achieving bimanual dexterity with RL + Sim2Real!
toruowo.github.io/bimanual-t…
TLDR - We train two robot hands to twist bottle lids using deep RL followed by sim-to-real. A single policy trained with simple simulated bottles can generalize to drastically different real-world objects.
Lin Shao retweeted
🚀RobotEra & @Tsinghua_Uni launch Humanoid-Gym! An open-source sim2real RL framework for humanoid robots. This codebase is verified on RobotEra's humanoid robot XBot (XiaoXing)!
Project page: sites.google.com/view/humano…
Github : github.com/roboterax/humanoi…
@JianyuChen_THU @wangyenjen
Lin Shao retweeted
Replying to @corl_conf
@corl_conf website and call for papers are live now!
corl.org/
Deadline is on June 6!
Looking forward to seeing everyone in Munich this November!
#CoRL2024
Lin Shao retweeted
Are a senior PhD student or early career researcher? It is time to check and apply for #RSSPioneers2024, held jointly with #RSS2024 conference!
The application portal for the #RSSPioneers2024 workshop (to be held at #RSS2024) is now open! 🤖 Senior Ph.D. students and early career researchers in robotics are encouraged to apply. Check out our website for further instructions: sites.google.com/view/rsspio…
Lin Shao retweeted
After a fruitful search for a faculty position in Machine Learning + Robotics, happy to share that I am joining the University of Cambridge.
Interested in working together? Reach out to connect!
(PhD applications due Dec 5th, MPhil - later in Spring)
youtube.com/watch?v=LlCwHnp3…
Lin Shao retweeted
Thanks to @ROBOTIS, one can now order all the LEAP Hand parts in a bundle (1-click) from robotis website:
=> LEAP parts: robotis.us/leap-hand-bundle-…
=>LEAP LITE parts: robotis.us/leap-hand-bundle-…
=> LEAP Hand design, code, API, sim2real: leaphand.com/
xitter.cf/pathak2206/status/1705…
Great to run into Sam and others from Robotis at IROS! LEAP hand parts can be bought with a single click from their website (robotis.us/leap-hand-bundle-…).
None of this would be possible without @kenny__shaw who couldn't attend but was there with us in spirit :)
Lin Shao retweeted
#ICRA got upgraded to rank A* (maximum rank) in the CORE ranking, well deserved! 👏👏👏 @ieee_ras_icra
Lin Shao retweeted
🤖 Introducing the future of cleanliness: an artificial intelligence-assisted cleaning robot! Wave goodbye to tedious chores and let cutting-edge AI technology handle the mess #FutureOfWork #AI – at Laconia, NH
Lin Shao retweeted
Happy to share our #ICCV2023 work by @weijiawu7 that turns stable diffusion into surprisedly good segmentation model for open-vocab words, like Eiffel Tower or Ultraman
Paper: arxiv.org/abs/2303.11681
Code: weijiawu.github.io/Diffusion…
"📚 Excited to share our work presented at ICCV23! We explored to leverage free cross-attention masks from Stable Diffusion for semantic segmentation labeling. The results not being very impressive, but intriguing. weijiawu.github.io/Diffusion… #ICCV23 #SemanticSegmentation"
Lin Shao retweeted
Congratulations to @GuibasLeonidas and colleagues for winning the SIGGRAPH 2023 Test of Time award for their paper “Functional Maps: A Flexible Representation of Maps Between Shapes”!
dl.acm.org/doi/10.1145/21855…
people.csail.mit.edu/jsolomo…
Lin Shao retweeted
Many physical systems are high-dimensional, but we only really care about some low-dimensional subspace.
Our latest work shows how to fit these subspaces as small neural maps automatically, *without* any data as input, just the energy function.
Read on to learn how! (1/N) 🧵
Lin Shao retweeted
If you are still around, I will be speaking at today's RSS Workshop on Learning Dexterous Manipulation on some cool work we are doing at @berkeley_ai! Room 323 at 4:45pm @RoboticsSciSys




































