Suning Huang
@suning_huang
PhD @Stanford｜BEng @Tsinghua_Uni. Learning to teach robots to learn. Nice to meet you ;)
Unfortunately I cannot attend the conference in person this year, but our co-author @Kevin_GuoweiXu will be presenting the paper and answer all your questions! 📜Poster session: Time: Wed 16 Jul 11 a.m. PDT — 1:30 p.m. PDT Location: West Exhibition Hall B2-B3 #W-607
🚀 Introducing MENTOR: Mixture-of-Experts Network with Task-Oriented Perturbation for Visual Reinforcement Learning! 🌟 We propose a strong model-free visual RL algorithm that can learn robust visuomotor policies from scratch – in the real world! 💪🤖 🌐 Check out the project…
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:…
What makes data “good” for robot learning? We argue: it’s the data that drives closed-loop policy success! Introducing CUPID 💘, a method that curates demonstrations not by "quality" or appearance, but by how they influence policy behavior, using influence functions. (1/6)
How can we move beyond static-arm lab setups and learn robot policies in our messy homes? We introduce HoMeR, an imitation learning agent for in-the-wild mobile manipulation. 🧵1/8
Introducing Mobi-π: Mobilizing Your Robot Learning Policy. Our method: ✈️ enables flexible mobile skill chaining 🪶 without requiring additional policy training data 🏠 while scaling to unseen scenes 🧵↓
🧑🤖 Introducing Human2Sim2Robot! 💪🦾 Learn robust dexterous manipulation policies from just one human RGB-D video. Our Real→Sim→Real framework crosses the human-robot embodiment gap using RL in simulation. #Robotics #DexterousManipulation #Sim2Real 🧵1/7
Excited to share that MENTOR has been accepted to #ICML2025! See you in Vancouver this July🤖 x.com/suning_huang/s…
🚀 Introducing MENTOR: Mixture-of-Experts Network with Task-Oriented Perturbation for Visual Reinforcement Learning! 🌟 We propose a strong model-free visual RL algorithm that can learn robust visuomotor policies from scratch – in the real world! 💪🤖 🌐 Check out the project…
𝑫𝒆𝒎𝒐𝑮𝒆𝒏 has been accepted to #RSS2025 🥳 See you in LA this June 🙌
Diffusion policies are powerful, but often require ~100 or more demos for effective performance. Can we save the labor and train policies with 𝐨𝐧𝐥𝐲 𝐨𝐧𝐞 human-collected demo? Check out 𝑫𝒆𝒎𝒐𝑮𝒆𝒏, a method for synthetic demo generation ✨ demo-generation.github.io (1/n)
When I was a Ph.D. student at @Caltech, @lschmidt3 discussed the paper "Do ImageNet Classifiers Generalize to ImageNet?" in his job talk, which left me with a super deep impression until today. Basically, they recreated an ImageNet and found the SOTA models in circa 2019 had…
No more trust any RL conclusions from mujoco benchmarks, especially about exploration. Many results are there simply because: 1) not using massive parallel simulation for exploration, 2) not correctly randomizing environments and goals, 3) not reasonably setting up partial obs.
🍌We present DenseMatcher！ 🤖️DenseMatcher enables robots to acquire generalizable skills across diverse object categories by only seeing one demo, by finding correspondences between 3D objects even with different types, shapes, and appearances.
🚀 Introducing LLaVA-o1: The first visual language model capable of spontaneous, systematic reasoning, similar to GPT-o1! 🔍 🎯Our 11B model outperforms Gemini-1.5-pro, GPT-4o-mini, and Llama-3.2-90B-Vision-Instruct! 🔑The key is training on structured data and a novel inference…
We released RoboDuet, a novel framework which employs two collaborative policies to realize locomotion and manipulation simultaneously. Policies trained by RoboDuet enable zero-shot deployment across different quadruped robots in the real world.
🚀 Excited to introduce Stem-Ob, A powerful, plug-and-play upgrade for visual imitation learning! Stem-OB makes your VIL policy robust to real-world visual disturbances like lighting and appearance changes, boosting performance with minimal cost! #Robotics #AI
The Stanford Robotics Center just launched last weekend, and we demonstrated dual robot bed making with EquiBot along with other demos in the domestic suite. We will present EquiBot at #CoRL2024 in Poster Session 1 (11/6 11AM). Please come to learn more. equi-bot.github.io
What makes visual representations truly effective for robotics? Introducing Manipulation Centricity that bridges visual representations & manipulation performance, leading to a simple yet powerful representation trained on large-scale robotic datasets. 👉: robots-pretrain-robots.github.io