
Junfeng Long retweeted
Reward is always a bottleneck for robotic policy finetuning. Dense rewards introduce bias or require laborious annotations, while still struggling with behavioral nuances. How to achieve dense rewards with minimal annotations at test time? hire-project.github.io #CoRL2026
Junfeng Long retweeted
Wanted to get into manipulation. Decided to first build a full stack that scales.
Plan: copy @GeneralistAI's gripper. Fun weekend project.
It was not.
So @t_k_233 and I documented every design choice and lesson along the way.
Meet Specialist: fully open-source hardware, data pipeline, training, and deployment code.
specialist-ai.github.io/
Junfeng Long retweeted
Can you believe these are open-loop rollouts?
Excited to introduce CARD 🃏 — Contact-Anchored Retargeting for Dexterous Manipulation.
Almost every method today is limited to kinematics alone as the 𝐢𝐧𝐭𝐞𝐫𝐟𝐚𝐜𝐞 between the human and the robot. With CARD, we track real world contact and forces alongside kinematics to perform unprecedented dexterous tasks through tactile reactive replay!
CARD is scalable:
📹 Multiple input sources
🤖 Multiple robot hands
⚡ needs no object models, no simulator, 10 Hz
Junfeng Long retweeted
The exponential time hypothesis is FALSE!
We are announcing three new discoveries in mathematics, computer science, and physics about harmonic analysis, complexity theory, and quantum field theory.
openai.com/index/building-sc…
Junfeng Long retweeted
It is a new paradigm -- world model + APG. This enables 1000x faster training of generalist whole-body tracker. And opus 5.5 high can reproduce it within 1h for lafan.
Recently I wonder if we are turning dumb and falling into scaling slop. We need more @qiayuanliao .
We found a way to scale humanoid control that reduces GPU-hours by more than 1000×.
Across 129,514 motions, SuperMimic surpasses the state of the art (128 GPUs, 7 days) with 1 GPU in 3 hours.
Project: supermimic.github.io
Demo: supermimic.github.io/demo/
Junfeng Long retweeted
Can you sample a robot policy the way you sample an image? 🎲
No RL run. No reward. No fine-tuning.
We train a diffusion model on the 𝘄𝗲𝗶𝗴𝗵𝘁𝘀 of thousands of RL policy networks, then sample brand-new ones that walk 🚶, dance 🕺, and spin-kick 🦵 on a real Unitree G1 🤖
The best part? The 𝘄𝗲𝗶𝗴𝗵𝘁𝘀 have a secret 👀 Every network, no matter the motion, pays the most attention to the hips and waist — embodiment comes first; motion comes second. 🦾
📄 arXiv: arxiv.org/abs/2610.20606
🌐 Website: humanoid-weight-diffusion.gi…
#Robotics #HumanoidRobots #DiffusionModels #RobotLearning #EmbodiedAI
Junfeng Long retweeted
Practice makes perfect. A truly agentic robot shouldn’t just know what to do. It should also figure out how to get better.
What if a robot could watch a demonstration, figure out what it needs to practice, and learn how to get better through practice?
Introducing RPG: Guided Self-Improvement for Embodied Agents, a real-to-sim-to-real pipeline that turns real-world demonstrations into targeted practice.
RPG reconstructs practice tasks from demonstrations, uses demo videos to guide skill improvement, and continually updates a shared skill system from failures.
On real robots, RPG achieves 30/30 successes across 3 tasks. In simulation, it reaches 95.0% success across 22 manipulation tasks after 15 rounds of self-improvement.
Demo → Reconstruct → Guided Practice → Self-Improve → Go Real
Can a humanoid learn robust whole-body loco-manipulation directly from a human? No robot, no teleop, no retargeting.
Meet Workhorse 🐴: our G1 topples and climbs a suitcase, catches a thrown box, and sorts boxes with hands and a kick. Fully autonomous.
More videos at hybridrobotics.github.io/wor…
Junfeng Long retweeted
Replying to @xiaolonw @chris_j_paxton
Any online optimization (IK, ID, MPC) for low-level control should be abandoned. A RL policy can fundamentally track more precisely, more robustly, and adaptively to unknown disturbances and uncharacterized noise, especially for low-cost hardware, even on a manipulator.
Junfeng Long retweeted
sharing our new project Smash!🏓
We built the first outdoor humanoid table tennis player with fully onboard perception — no MoCap, no external cameras. check the video for details.
paper and code on the way!
Junfeng Long retweeted
Happy to see my code could solve the problems my labmate previously refused to tackle until I made it easy.
Funny how it turned into a scoop of my own PhD thesis topic, but glad the tools work. 🙃
Junfeng Long retweeted
Can a single learned controller generalize across diverse humanoid embodiments?
Introducing XHugWBC, a novel cross embodiment training framework that enables generalist humanoid control through:
1) physics-consistent morphological randomization
2) unified state-action representation with semantic alignment across different robots
3) graph-based policy for cross-humanoid control
We find that a single policy can zero-shot generalize to unseen robots with one-time training. The resulting generalist policy reaches approximately 85% of the performance achieved by the specialist, and the fine-tuning generalist shows approximate 10% improvement compared to the generalist policy.
🔗Website:xhugwbc.github.io
📕 Arxiv:arxiv.org/abs/2602.05791
Junfeng Long retweeted
💡 How can humanoids learn adaptable skills from a single human motion?
🤖 Introducing AdaMimic: Towards Adaptable Humanoid Control via Adaptive Motion Tracking
Paper: arxiv.org/pdf/2510.14454
Website: taohuang13.github.io/adamimi…
Junfeng Long retweeted
How do we learn motor skills directly in the real world? Think about learning to ride a bike—parents might be there to give you hands-on guidance.🚲
Can we apply this same idea to robots?
Introducing Robot-Trains-Robot (RTR): a new framework for real-world humanoid learning.
Junfeng Long retweeted
Humanoid robots should not be black boxes 🔒 or budget-busters 💸!
Meet Berkeley Humanoid Lite!
▹ 100% open source & under $5k
▹ Prints on entry-level 3D printers—break it? fix it!
▹ Modular cycloidal-gear actuators—hack & customize towards your own need
▹ Off-the-shelf parts—NO custom PCBs, NO CNCs
Built on 4 years of R&D to put humanoid robotics in everyone's reach.
Website: lite.berkeley-humanoid.org
Documentation: berkeley-humanoid-lite.gitbo…
Junfeng Long retweeted
Atlas is demonstrating reinforcement learning policies developed using a motion capture suit. This demonstration was developed in partnership with Boston Dynamics and @rai_inst.
Junfeng Long retweeted
💡Can a humanoid robot learn to stand up across diverse real-world scenarios from scratch?
🤖 Introducing HoST: Learning Humanoid Standing-up Control across Diverse Postures
Website: taohuang13.github.io/humanoi…
Junfeng Long retweeted
🚀 Can we make a humanoid move like Cristiano Ronaldo, LeBron James and Kobe Byrant?
YES!
🤖 Introducing ASAP: Aligning Simulation and Real-World Physics for Learning Agile Humanoid Whole-Body Skills
Website: agile.human2humanoid.com/
Code: github.com/LeCAR-Lab/ASAP
Junfeng Long retweeted
Excited to introduce the Perceptive Internal Model (PIM) for Humanoid Robots!
The first policy simultaneously for:
- Go up and down stairs, jump gaps, and 50cm high platforms.
- Indoor and outdoor scenarios.
- Unitree H1 and Fourier GR-1 robots.
Paper: arxiv.org/abs/2411.14386
Junfeng Long retweeted
Glad that Learning H-Infinity Locomotion Control receives the best poster award at the #CoRL2024 workshop “LocoLearn: From Bioinspired Gait Generation to Active Perception”!






















