
Project title: 도시형 생성 지능을 위한 사회적 군중 시뮬레이션 기술 개발 (H100 × 8) Project title: 희소 궤적 월드 액션 모델을 활용한 양손 조작 행동 생성 기술 개발 (B200 × 8)

The paper is “Unified Prediction and Planning via Conflict-Aware Disjoint Parameter Training” by Taewon Seo*, Seonae Jeon*, Giwon Lee*, Kukjin Yoon† and Daehee Park†

Project title: 다중 휴머노이드의 자율 협업을 위한 비전-언어 모델 기반 적응형 장기·복합작업 계획 기술 Project title: 아카이브 영상 콘텐츠 활용을 위한 AI Outpainting 및 콘텐츠 변환 기술 개발 Project title: 한국형 Physical LLM 의 실시간성과 안정성을 위한 가변형 추론 프레임워크 개발

Project title: 자기 진화형 Physical LLM: 불확실한 환경에서 행동 파운데이션 모델의 강건하고 효율적인 적응 기술 연구 News: Link

(1) “EggHand: A Multimodal Foundation Model for Egocentric Hand Pose Forecasting” by Jaeyoung Choi*, Hyeondong Kim*, Yujin Kim and Daehee Park (2) “BitTP: The Lightweight Trajectory Prediction Model with BitLLM for Edge-Devices” by Mincheol Kang*, Hyunjin Lim*, Bomin Kang* and Daehee Park Congratulations to all authors 🎉

The paper is “CascadeOcc: Rethinking 3D Occupancy World Models with Cascaded VQ Representations ” by Kyumin Hwang*, Wonhyeok Choi*, Jaeyeul Kim, Jihun Park, Daehee Park† and Sunghoon Im†

Six new master’s students have joined the lab. Hyungchul Shin joins from the Department of Electrical Engineering and Computer Science (EECS), and Youngjin Park, Hyunjin Lim, Sihyeong Lee, Hyeonwoo Park, Donggyu Lee from the Department of Artificial Intelligence (AI). Welcome to ISL Lab!

Project title: Extending Action Foundation Models to Dynamic Human-Robot Collaboration, for inclusion in our Academic Grant Program

Project title: 행동 파운데이션 모델의 동적 환경으로 확장을 통한 안전하고 범용적인 인간-로봇 협동 연구 News: Link

Two new master’s students have joined the lab. Jaeyoung Choi joins from the Department of Electrical Engineering and Computer Science (EECS), and Taewon Seo from the Department of Artificial Intelligence (AI). Welcome to ISL Lab!

Project title: Physical AI 분야의 거대 생성모델 기술 선도를 위한 LG AI STAR 인재양성 사업 News: Link

(1) “Generative Active Learning for Long-tail Trajectory Prediction via Controllable Diffusion Model” by Daehee Park, Monu Surana, Pranav Desai, Ashish Mehta, Reuben MV John, Kuk-Jin Yoon , (2) "Interaction-Merged Motion Planning: Effectively Leveraging Diverse Motion Datasets for Robust Planning" by Giwon Lee*, Wooseong Jeong*, Daehee Park, Jaewoo Jeong, Kuk-Jin Yoon (Highlight ✨).

The paper is “Non-differentiable Reward Optimization for Diffusion-based Autonomous Motion Planning ” by Giwon Lee*, Daehee Park*, Jaewoo Jeong*, Kuk-Jin Yoon

Project title: 인간 중심 공생적 임바디드 인공지능 시스템. News: Link

The paper is “Multi-modal Knowledge Distillation-based Human Trajectory Forecasting” by Jaewoo Jeong, SeoHee Lee, Daehee Park, Giwon Lee, and Kuk-Jin Yoon.

Intelligent Systems and Learning Laboratory at DGIST EECS has launched in Feb 2025 by Prof. Daehee Park.