Funded research
Research Projects
Research and development projects since 2018
- Ongoing projects
- 16
- Led as PI
- 12
- Grant programmes
- 19
Ongoing Projects
Reconfigurable Lifelong Learning of VLA Models for Super-adaptive Embodied Agents
초적응 체화 에이전트를 위한 VLA 모델의 재구성형 평생학습 기법 연구
Using B200 GPU computing support to study reconfigurable lifelong learning of vision-language-action models for highly adaptive embodied agents, including iterative experience collection, curation, retraining, and rollout evaluation.
2026.11.20 (4 months)
Principal Investigator
Test-Time Storage Scaling for LLM Serving (HiPER Phase 3)
LLM Serving 최적화를 위한 Test-time Storage Scaling 기술 연구 (HiPER 3기)
Investigating test-time storage scaling that combines neural memory and vector storage to improve long-context handling, reasoning, and inference efficiency in large language model serving.
2026.07.01 – 2029.06.30
Resource-Intensive AI Technologies Based on Sustainable GPU Integrated Platforms
지속 가능한 고집적 GPU 인프라 기반의 자원 집약적 AI 원천기술 개발
Integrating distributed university GPU resources into shared computing infrastructure through the IITP computing resource-intensive AI application technology development program. The project develops GPU pooling and GPU-as-a-Service operations, with the group's research focusing on physical AI foundation models and collaboration among heterogeneous robots.
2026.04.01 – 2028.12.31
Sub-project Investigator
Development of Leading Talent in Medical Domain-Specific Generative AI
의료특화 생성AI선도인재양성
Training researchers through industry-academic collaboration on medical generative AI, including adaptive state estimation, vision-language-action skills, and safe physical AI control for medical applications.
2026.04.01 – 2029.12.31
Participant
V2X Operation Optimization Using Deep Reinforcement Learning
V2X 운영 최적화를 위한 강화학습 기반 기술 개발
Developing reinforcement learning methods for electric vehicle charging, discharging, and energy-market bidding, with scalable fleet scheduling and robust adaptation to uncertain operating conditions.
2026.03.30 – 2027.03.29
Principal Investigator
Reactive Skill Diffusion for Real-world Embodied AGI
실세계 범용 체화 지능 실현을 위한 반응형 스킬 디퓨전 연구
Developing reactive skill diffusion guided by world models, collaborating agents, and memory to support real-world adaptation, heterogeneous robot collaboration, and continual growth of embodied intelligence.
2026.03.01 – 2029.02.28
Principal Investigator
Retrieval-Augmented Knowledge Search with LG Electronics
LG RAG: 연구개발 문서 기반 지식검색 고도화
Improving retrieval-augmented knowledge search for research and development documents through document-level context, query decomposition, and knowledge graph integration.
2026
Autonomous Cluster Management
Autonomous Cluster Management 기술 (삼성 RCA)
Developing a multi-agent framework for GPU cluster fault detection, root cause analysis, and self-healing, combining operational data analysis with the generation and verification of recovery policies.
2025.12.30 – 2026.10.29
Principal Investigator
AI Star Fellowship Support Program(Sungkyunkwan Univ.)
AI스타펠로우십지원(성균관대학교)
Developing synergistic multi-agent technologies for companion AI services, including collaborative planning, interoperability, continual collective learning, and well-care service integration. The lab leads subproject 2 within the Sungkyunkwan University AI Star Fellowship consortium.
2025.07.01 - 2030.12.31
Sub-project Investigator
Accelerated Insight Reasoning via Continual Learning
지속학습을 통한 인사이트 고속 추론 연구
Developing advanced techniques for policy adaptation in reinforcement learning using cross-domain skill diffusion methods for improved generalization
2025.04.01 - 2029.12.31
Principal Investigator
Development of Open-Ended Alignment Fluxional AI for Ever-changing Environment and Value
변화하는 환경과 가치에 지속 부합하는 Open-ended Alignment 인공지능 기술 개발
Contributing to Out-thinking Alignment AI within the Open-Ended Alignment consortium, with research on semantic-skill-based learning, adaptation to changing environments, and replanning for embodied agents.
2024.07.01 - 2031.12.31
Participating Investigator
BK21 Education and Research Center for Sustainable Future Computing
[BK21] 지속성장 미래컴퓨팅 교육연구단
Supporting graduate students and research training in software and AI through the laboratory's participation in the BK21 program.
2024.03.01 – 2027.08.31
Participant
Self-directed Multi-modal Intelligence for solving unknown, open domain problems
오픈 도메인 멀티모달 자기주도 인공지능 기술 개발
Securing autonomous learning and adaptation technology for open-domain complex problem situations and demonstrating platform-based agent services Advancing self-supervised learning, meta few-shot learning, and lifelong reinforcement learning for autonomous AI and developing integrated models
2022.04.01 - 2026.12.31
Principal Investigator
Adaptive Personality for Intelligent Agents
개성 형성이 가능한 에이전트 플랫폼 기술 개발
Development of AI agent technology that forms personality based on multimodal information of the target, learns according to the interaction partner, and exhibits social behavior
2022.04.01 - 2026.12.31
Sub-project Investigator
ICT Creative Consilience Program
ICT명품인재양성사업(성균관대)
Cultivating innovative leader-type professional researchers to solve future societal problems and lead new technologies based on ICT technology and multidisciplinary knowledge
2020.07.01 – 2029.12.31
Participant
Artificial Intelligence Graduate School Program(Sungkyunkwan University)
인공지능대학원 지원사업 (성균관대학교)
Supporting graduate education and research in artificial intelligence at Sungkyunkwan University and training researchers through AI research and academic activities.
2019.04.01 – 2028.12.31
Participant
Completed Projects
AI Computing Infrastructure Enhancement (GPU Rental Support)
AI컴퓨팅자원활용기반강화(GPU 임차 지원) 사용자 지원사업
Supporting AI research with GPU computing resources through the MSIT GPU Rental Support User Support Program (RQT-25-120157), including language-model-based skill grounding and embodied-agent research.
2026.01.01 – 2026.06.30
Principal Investigator
Development of Medical Domain-Specific Foundation Models with Acryl
의료특화 파운데이션 모델 개발 (아크릴)
Developing foundation models for medical applications through an industry-academic research project with Acryl.
2025.05.01 – 2025.11.30
Principal Investigator
Adaptive Privacy Compliance with Hybrid Legal AI and Open-Weight Language Models
규칙기반 하이브리드 법률 AI와 오픈웨이트 기반 sLLM 모델을 활용한 실시간 적응형 개인정보 컴플라이언스 솔루션 개발
Combining rule-based expert systems and open-weight language models to check privacy requirements, verify compliance checklists, and assist with document corrections.
2024.09.01 – 2026.08.31
Principal Investigator
AI-Based PS-LTE Access Optimization with PolarisAI
재난안전통신망과 철도망 전파간섭 지역에서 AI기반 대규모 PS-LTE 단말기 접속방법 개선에 대한 연구
Researching AI-based methods to improve large-scale PS-LTE terminal access in areas where public safety and railway networks experience radio interference, in collaboration with PolarisAI.
2024.08.01 – 2024.12.31
Principal Investigator
HiPER (High-Performance Exabyte stoRage)
대규모 AI 지향 Peta-Scale 호스트-스토리지 Co-Design
Focusing on large-scale AI workloads, we aim to develop a host software stack optimized for storage access patterns, design a storage architecture tailored to these patterns, and further optimize the integrated host software stack for efficient Peta-scale data processing.
2024.05.16 – 2026.05.15
Development of an Open-Source LLM-Based Knowledge Graph Generation Pipeline
오픈소스 LLM 기반 자체 Knowledge Graph Generation Pipeline 개발
Design and implementation of a pipeline to automate knowledge graph generation using LLMs, focusing on prompt design and application testing for tasks such as NER and RE. The project includes benchmarking open-source LLMs, dataset selection, text preprocessing, prompt engineering, and evaluation of the generated knowledge graphs.
2023.09.01 – 2023.11.30
Distributed Learning for Elastic Scaling with Samsung Electronics
탄력적 스케일링을 위한 분산학습 기술
Developing distributed learning scheduling algorithms and scalable evaluation models for systems of varying sizes, in collaboration with Samsung Electronics.
2023.06.16 – 2024.04.15
Policy Generalization via Multimodal Skill Transfer
멀티모달 스킬 전이 기반 정책 일반화 기법 연구
Targeting real-world environments (dynamic environments and complex multitasks) where Embodied AI agents operate, we aim to achieve a high level of policy generalization for agents capable of quickly learning and adapting problem-solving methods for new tasks (unseen tasks) and unexplored, dynamic environmental conditions (unexplored, dynamic environments)
2023.03.01 – 2026.02.28
Principal Investigator
Detection Method for Factors Influencing AI Agent Quality
AI 에이전트 품질 영향 요소 검출 기법
Research on methods to analyze and improve the decision-making quality of AI agents by identifying influential policy factors, diagnosing service errors, and supporting effective policy network debugging in complex environments.
2022.05.16 – 2022.12.16
Elastic GPU Cluster Scheduling and Distributed Learning Management
탄력적 스케일링을 위한 분산학습 관리기술 연구 (GPU 클러스터 탄력적 스케줄링 및 배치)
Developing elastic distributed learning management to improve GPU cluster utilization and reduce training time, with cluster monitoring, scaling-overhead optimization, and management agents in collaboration with Samsung Electronics.
2022.04.01 – 2022.12.31
Development of a Personal Online Golf Simulator (EXTRACE) Based on GIS Information and AI Technology
GIS 정보와 AI기술 기반 개인용 온라인 골프 시뮬레이터(EXTRACE) 개발
Development of a personal golf simulator Securing technology for the development of a personal golf launch monitor with expert-level measurement accuracy
2021.09.01 – 2023.08.31
Sub-project Investigator
Development of a framework to facilitate applications and developments of large-scale AI between a datacenter and an edge NPU based on federated inference/training
데이터센터-엣지 NPU 간 연합 추론/학습 기반 대규모 인공지능 응용/개발을 용이하게 하는 SW 프레임워크 개발
Advancement of federated learning processors between edge and server Development of federated inference processors between edge and server Development of edge NPU hardware performance simulator/profiler Development of distributed deployment algorithms for federated inference models across edge-server NPUs
2021.04.01 – 2024.12.31
Sub-project Investigator
Adaptive Federated Learning in Dynamic Heterogeneous Environment
동적인 디바이스 환경에서 적응적 연합학습기술 개발
Development of adaptive federated learning technology that is robust to statistical heterogeneity of data and system heterogeneity of devices in dynamic device and data environments, while optimally reflecting the characteristics of individual devices and users as well as application requirements
2021.04.01 – 2022.12.31
Sub-project Investigator
Deep Learning-Based 3D Golf Swing Analysis and Launch Monitor (EXSWING)
딥러닝 기반 3D 스윙 분석 골프 론치모니터(EXSWING) 개발
Developing a golf launch monitor that uses deep learning and dual-camera 3D pose estimation to analyze swings, recognize clubs, and provide personalized practice guidance.
2020.07.01 – 2021.06.30
Participant
Development of Core AI Software for Drones
드론을 위한 AI 핵심 S/W개발
Development and continuous improvement of core AI technologies and software for services such as situation understanding, autonomous control, swarm collaboration, and emergency response using large-scale data collected from swarm drones, with public release
2020.04.23 – 2024.12.31
Sub-project Investigator
Research on Storage Systems and Application Framework Optimization for Next-Generation Computing Environments
차세대 컴퓨팅환경을 위한 스토리지시스템 및 응용프레임워크 최적화 기술 연구
Research on data-centric computing environments leveraging SmartSSDs extended with CPUs, FPGAs, and GPUs, focusing on system software optimization for data center-scale storage systems based on NVMe-oF, as well as application-level optimization for DBMS and distributed key-value systems utilizing next-generation SSDs.
2019.08.01 – 2022.07.31
Real-Time Data Management and Adaptive Model Serving for Autonomous Systems
자동화 시스템을 위한 실시간 데이터 관리와 적응적 모델 서빙
An adaptive model serving architecture based on real-time data management that executes appropriate machine learning models and makes decisions without delay under limited resource conditions to proactively respond to changes in the physical operating environment of automated systems
2018.11.01 – 2020.10.31
Principal Investigator