Kim Dongjin

I am an integrated M.S.–Ph.D. student in the DAVIAN Lab at KAIST AI, advised by Professor Jaegul Choo. My research centers on Vision-Language-Action (VLA) models for robotics.

This focus grew out of a single thread running through my earlier work. I spent several years on multi-modal grounding — localizing sound sources in visual scenes and understanding video — learning to align perception across modalities in space and time, and on language grounding through LLM safety and information extraction. VLA is where these threads converge: grounding both perception and language to drive action in the real world.

Research Interests

Focus

Vision-Language-Action (VLA)
Robot Manipulation

Multimodal Grounding

Audio-Visual Learning
Video Understanding
Sound Source Localization

Language Grounding & Safety

LLM Safety
Information Extraction / RAG

News

Jun 2026 3D HAMSTER — a hierarchical Vision-Language-Action model with 3D trajectory guidance for robot manipulation — accepted to IROS 2026 (co-first author). Presenting in late Sep 2026.
Apr 2026 Two papers presented at ICLR 2026 — ExpGuard (co-first author) and LiveWeb-IE.
Jun 2025 Presented Object-aware Sound Source Localization via Audio-Visual Scene Understanding at CVPR 2025, Nashville, TN.
Mar 2025 Started Integrated M.S.–Ph.D. program at KAIST AI, DAVIAN Lab, advised by Prof. Jaegul Choo.
Feb 2025 Paper accepted at AAAI 2025 — Watch Video, Catch Keyword: Context-aware Keyword Attention for Moment Retrieval and Highlight Detection.
Jun 2024 Presented Learning to Visually Localize Sound Sources from Mixtures without Prior Source Knowledge at CVPR 2024, Seattle, WA.

Publications

2026

No image
3D HAMSTER: Bridging Planning and Control in Hierarchical Vision Language Action Models through 3D Trajectory Guidance

Dongyoon Hwang*, Byungkun Lee*, Dongjin Kim*, Hyojin Jang, Hoiyeong Jin, Jueun Mun, Minho Park, Hojoon Lee, Hyunseung Kim, Jaegul Choo (*Equal contribution)
IROS 2026
- #VLA
- #Robot Manipulation
- #3D Trajectory
[project] [arXiv] [code] [model]
No image
ExpGuard: LLM Content Moderation in Specialized Domains

Minseok Choi*, Dongjin Kim*, Seungbin Yang, Subin Kim, Youngjun Kwak, Juyoung Oh, Jaegul Choo, Jungmin Son (*Equal contribution)
ICLR 2026
- #LLM Safety
- #Content Moderation
- #Specialized Domains
[arXiv] [OpenReview] [code]
No image
LiveWeb-IE: A Benchmark For Online Web Information Extraction

Seungbin Yang, Jihwan Kim, Jaemin Choi, Dongjin Kim, Soyoung Yang, ChaeHun Park, Jaegul Choo
ICLR 2026
- #Benchmark
- #Information Extraction
- #Web
[arXiv] [OpenReview]

2025

No image
Object-aware Sound Source Localization via Audio-Visual Scene Understanding

Sung Jin Um*, Dongjin Kim*, Sangmin Lee, Jung Uk Kim (*Equal contribution)
CVPR 2025
- #Multi-modal
- #Video Understanding
- #Sound Source Localization
[arXiv] [code]
No image
Watch Video, Catch Keyword: Context-aware Keyword Attention for Moment Retrieval and Highlight Detection

Sung Jin Um, Dongjin Kim, Sangmin Lee, Jung Uk Kim
AAAI 2025
- #Multi-modal
- #Video Understanding
- #Moment Retrieval
- #Highlight Detection
[arXiv] [code]

2024

No image
Learning to Visually Localize Sound Sources from Mixtures without Prior Source Knowledge

Dongjin Kim*, Sung Jin Um*, Sangmin Lee, Jung Uk Kim (*Equal contribution)
CVPR 2024
- #Multi-modal
- #Video Understanding
- #Sound Source Localization
[arXiv] [code]

2023

No image
Audio-Visual Spatial Integration and Recursive Attention for Robust Sound Source Localization

Sung Jin Um*, Dongjin Kim*, Jung Uk Kim (*Equal contribution)
ACM MM 2023
- #Multi-modal
- #Video Understanding
- #Sound Source Localization
[arXiv] [code]

Education

2025–Present

Integrated M.S.–Ph.D. in Artificial Intelligence

KAIST AI · DAVIAN Lab

Advisor: Prof. Jaegul Choo

Research focus: Vision-Language-Action, multimodal grounding, language grounding & safety
2019–2025

B.S. in Computer Science & Engineering

Kyung Hee University · Visual AI Lab

Advisor: Prof. Jung Uk Kim

Research focus: Multi-modal learning, Video Understanding, Sound Source Localization

Experience

2025.03–Present

Graduate Researcher

KAIST AI · DAVIAN Lab

Research on VLA, LLM Safety, and multi-modal learning.
2024.03–2024.08

AI Engineer

LETSUR (AI Startup)

Development of AI-based services utilizing LLMs and RAG.
2024.01–2024.02

AI Research Intern

ETRI (Electronics and Telecommunications Research Institute)

Research on Video Moment Retrieval and Highlight Detection.
2022.11–2025.02

Undergraduate Researcher

Kyung Hee University · Visual AI Lab

Supervisor: Jung Uk Kim.

Talks & Presentations

IROS 2026

3D HAMSTER: Bridging Planning and Control in Hierarchical Vision Language Action Models through 3D Trajectory Guidance

Upcoming · IROS 2026 · Sep 2026
[project]
CVPR 2025

Object-aware Sound Source Localization via Audio-Visual Scene Understanding

Poster · Nashville, TN, USA · Jun 2025
[poster]
CVPR 2024

Learning to Visually Localize Sound Sources from Mixtures without Prior Source Knowledge

Poster · Seattle, WA, USA · Jun 2024
[poster]
ACM MM 2023

Audio-Visual Spatial Integration and Recursive Attention for Robust Sound Source Localization

Poster · Ottawa, Canada · Oct 2023
[poster]
Invited Talk 2025

"My Experience of CVPR Acceptance as an Undergraduate and Advice for Graduate School"

NexusAI Cohort 2 Final Seminar · Seoul, South Korea · 2025
[slides]