Minyoung Hwang @RSS 2025

Interested in how generative AI can be used for human-robot interaction? We’re organizing the 2nd Workshop on Generative AI for Human-Robot Interaction (GenAI-HRI) at #RSS2025 in LA — bringing together the world's leading experts in the field. The workshop is happening on Wed,…

Happening now at RTH 109!

How to generate billion-scale manipulation demonstrations easily? Let us leverage generative models! 🤖✨ We introduce Dex1B, a framework that generates 1 BILLION diverse dexterous hand demonstrations for both grasping 🖐️and articulation 💻 tasks using a simple C-VAE model.

Amazing to see tactile sensors that are 3d printable in any form factor!

“As a PHD student, your job is not publishing a paper every quarter. Focus on a problem in deep understanding and solve it in years under the protect of your adviser” from @RussTedrake #RSS2025

If you are at #RSS2025, check out our workshop on Generative AI for Human-Robot Interaction. We have a stacked lineup of speakers and panelists!

I’ll talk about the PARTNR framework and how LLMs perform in planning in dynamic environments. I’ll also talk about a unified memory architecture for robotics, an alternative to recent scene representations that rely on ad-hoc combination of multiple large models.

Most assistive robots live in labs. We want to change that. FEAST enables care recipients to personalize mealtime assistance in-the-wild, with minimal researcher intervention across diverse in-home scenarios. 🏆 Outstanding Paper & Systems Paper Finalist @RoboticsSciSys 🧵1/8

I’ll be giving a spotlight talk at the RSS SemRob workshop (OHE #122, 9:50-10am) about this work today! The talk is followed by the poster session, so feel free to stop by if you’re interested :) Happy to catch up or chat about research and potential collaboration during the…

What if an LLM could update its own weights? Meet SEAL🦭: a framework where LLMs generate their own training data (self-edits) to update their weights in response to new inputs. Self-editing is learned via RL, using the updated model’s downstream performance as reward.

Happening now! Drop by poster #168 at CVPR to see our work! Also giving a spotlight talk at CVPR EAI workshop at 3:50-4pm. Happy to chat w/ anyone interested during the conference😊

Gemini 2.0 can reason about the physical world! Try it out today at aistudio.google.com/starter-apps/s… Your robots will thank you for it :)

Excited to introduce 𝐋𝐀𝐏𝐀: the first unsupervised pretraining method for Vision-Language-Action models. Outperforms SOTA models trained with ground-truth actions 30x more efficient than conventional VLA pretraining 📝: arxiv.org/abs/2410.11758 🧵 1/9

This paper inspires me to not only selectively choose partial goals (demonstration, language, environment states like objects, etc.) for robot learning, but also to use a combination of them (e.g., demonstration + language)! Using CVAE and a learned prior seems to be the key🤔

Big shoutout to my awesome collaborators @JoeyHejna @DorsaSadigh @ybisk on this project😎 Thanks to @Hao_Zhu @DanielXieee @SoYeonTiffMin @lmathur @viddivj @_Yingshan @rosie_vitiello + all others in @CarnegieMellon CLAW lab for thoughtful feedback and help with data collection!👩‍💻