Haeone Lee @ ICML

Robot Learning Research @ KAIST

[Haeone Lee @ ICML @Haeone_Lee]

🤖 How can we learn a reliable policy across different robots and dynamics?

Excited to introduce SPACE, a framework that significantly improves cross-embodiment and cross-hardware (e.g., DROID) learning by addressing dynamics gaps, with execution-time adaptation.

📄 paper: https://t.co/zgByzVwFyz
📷 Project website: https://t.co/ZytiBGmMrY

🧵[1/n]
#Robotics #CrossEmbodiment .

[Jaehwi Song @j__aehwi]

Introducing HABIT — a large-scale robot manipulation dataset for human-present environments, where a person shares the workspace and interacts with the robot in every episode.

60 tasks · 10,563 episodes · 164 hours of rich human-robot interaction.

Toward robots that are not just capable, but safe and socially compatible around people.
https://t.co/kEtkqbuoIn

🧵[1/7]

[Haeone Lee @ ICML @Haeone_Lee]

📄 Paper: https://t.co/zgByzVwFyz
📷 Project page: https://t.co/ZytiBGmMrY

Special thanks to awesome collaborators
@bkjeon1211 @Suchaeck @jiankimr and adviser
@kimin_le2🙏.

P.S. I will be attending ICML 2026 in Seoul, Korea 🇰🇷. Feel free to reach out to me to chat about robot learning!
🧵[n/n]

[Haeone Lee @ ICML @Haeone_Lee]

🤖 How can we learn a reliable policy across different robots and dynamics?

Excited to introduce SPACE, a framework that significantly improves cross-embodiment and cross-hardware (e.g., DROID) learning by addressing dynamics gaps, with execution-time adaptation.

📄 paper: https://t.co/zgByzVwFyz
📷 Project website: https://t.co/ZytiBGmMrY

🧵[1/n]
#Robotics #CrossEmbodiment .

[Haeone Lee @ ICML @Haeone_Lee]

SPACE remains robust under dynamic shifts during deployment.

During training, it has only seen the empty box, and we put heavy metal into the box during inference.

SPACE adapts control commands to lift the heavy box, while vanilla policy fails.

It can also execute the policy under different control Hz and controller gains.

[JaeHyeok Doo @_jaehyeokdoo2]

Excited to introduce our work, Q-Flow: Stable and Expressive Reinforcement Learning with Flow-Based Policy, which has been accepted to ICML 2026!

By leveraging flow-consistent values, we resolve the critical trade-off between expressivity and stability in Flow-based Reinforcement Learning.

Joint work at KAIST w/ @bkjeon1211 , @SeonghyeonYe , @kimin_le2 , @seo_minjoon .

Paper: https://t.co/J5FR6ac9OF
Code: https://t.co/1tObfGord1
Project Page: https://t.co/BF0JWAKSKk

[Haeone Lee @ ICML @Haeone_Lee]

QoQ will be presented Today (Jun 4) at 15:00 - 16:30 in Hall C, #153.

Let's chat if you are interested in robot data curation or robot learning!

[Haeone Lee @ ICML @Haeone_Lee]

🚨Quality over Quantity in Robot Learning
Ever wondered why more robot trajectories did not lead to better policy performance?

We introduce QoQ, a robot data curation method that selects demonstrations for policy using influence functions.

We will present QoQ in ICRA 2026 on June 4. See you in Vienna!
📄 Paper: https://t.co/NY4EPzIopG

#ICRA2026 #RobotDataCuration #RobotLearning
[🧵1 / N] 👇

[Haeone Lee @ ICML @Haeone_Lee]

QoQ will be presented at ICRA 2026 during Interactive Session 6 on June 4, 2026.

Let's chat in Vienna, Austria🇦🇹!
📄 Paper: https://t.co/3DGZvbGaH1
💻 Code: https://t.co/2tbNGiOjMG

This work would not have been possible without my amazing co-authors. Thank you all once again!

[Haeone Lee @ ICML @Haeone_Lee]

🚨Quality over Quantity in Robot Learning
Ever wondered why more robot trajectories did not lead to better policy performance?

We introduce QoQ, a robot data curation method that selects demonstrations for policy using influence functions.

We will present QoQ in ICRA 2026 on June 4. See you in Vienna!
📄 Paper: https://t.co/NY4EPzIopG

#ICRA2026 #RobotDataCuration #RobotLearning
[🧵1 / N] 👇

[Haeone Lee @ ICML @Haeone_Lee]

Training policy by filtering data using QoQ scores improves policy success rate across multiple tasks in simulation and real robots.

QoQ can also utilize policy rollout as a validation set. This eliminates the need for separately collecting a validation set.

[Config @config_inc]

Hello world 🤖👋🏻—We are Config.
Today, we’re excited to share a preview (🔗 https://t.co/M6mnlt6waf) of what we’ve been building. Our mission is to make robots capable of reliably performing two-handed tasks across diverse real-world settings materially more cost- and time-efficient to deploy.

[Erdem Bıyık @ebiyik_]

Humanity's move 78 by @Ken_Goldberg