Development of LLM-based Safety and Reliability Enhancement Technologies for Open-Source Autonomous Driving SDVs
오픈소스 자율주행 SDV를 위한 LLM 기반 안전성 및 신뢰성 증강 기술 개발

Period: Mar 2024 – Mar 2030

Program (KR): 지역지능화혁신인재양성(경남대학교) 사업
Program (EN): Regional Intelligence Innovation Talent Development Program (Kyungnam University)

과제 내용 (KR):
- 실차 대비 1/2 크기의 Software Defined Vehicle(SDV) 제작
- 실차 대비 1/2 크기의 고속 자율주행 SDV 구현
- F1Tenth Simulator를 사용한 강화학습 기반의 고속 자율주행 기능의 최적화
- 정형검증 기반의 자율주행 안정성 검증

Project Scope (EN):
- Development of a half-scale Software Defined Vehicle (SDV) platform relative to a full-scale vehicle
- Implementation of a high-speed autonomous driving SDV at half-scale vehicle size
- Optimization of reinforcement-learning-based high-speed autonomous driving functions using the F1Tenth Simulator
- Verification of autonomous driving safety based on formal verification methods

Mathematical Quantification of Medical Diseases and Predictive Modeling for Clinical Treatment in Advanced Medical AI

Period: Mar 2023 – Feb 2027

Funding Agency: National Research Foundation of Korea (NRF), Mid-Career Researcher Support Program
Ministry: Ministry of Science and ICT (MSIT)
Role: Principal Investigator
Affiliation: Cyber Safety Lab, Gyeongsang National University (through Defense Industry Technology Convergence Research Institute)
Project No.: RS-2023-NR07679861382116530003
Total Duration: Mar 1, 2023 – Feb 28, 2027 (Current phase: Mar 1, 2025 – Feb 28, 2026)
Total Budget: ₩366,988,000 (Current Year: ₩91,747,000)
Legal Classification: National R&D Program (Innovation Act applicable), Security Level: Non-classified

Research Goal: Advancement of medical AI through mathematical quantification of disease characteristics and predictive modeling for clinical treatment planning.

Key Approaches: Integration of statistical modeling, data-driven learning, and AI-based decision frameworks to improve diagnostic accuracy and treatment outcome prediction.

AI-based Toothbrushing Guidance and Habit Improvement Algorithms

Period: Sep 2022 – Dec 2023 (Completed)

Program: Leaders in INdustry-university Cooperation (LINC 3.0)
Funding Agency: NRF / Ministry of Education
Role: Principal Investigator
Affiliation: Office of Leaders in Industry–University Cooperation, Gyeongsang National University
Combined Budget: ₩104,000,000
Legal Classification: Non-National R&D (Innovation Act applicable), Non-classified

Research Goal: To develop AI algorithms that analyze individual brushing behavior, improve oral hygiene habits, and provide personalized real-time guidance for microcurrent toothbrushes.

Key Outcomes: Built sensor- and motion-based models that detect brushing inefficiencies, analyze user behavior, and deliver adaptive feedback and recommendations for personalized oral care.

Network Slicing and Dynamic TSN for Autonomous Vehicle Edge Handover and Wireless Edge Systems

Period: Oct 2019 – Feb 2023 (Completed)

Related Support: IITP project on Wi-Fi network slicing and NRF mid-career project on dynamic TSN and network slicing for autonomous vehicle edge handover
Funding Agencies: Institute of Information & Communications Technology Planning & Evaluation (IITP); National Research Foundation of Korea (NRF) / Ministry of Science and ICT (MSIT)
Role: Principal Investigator
Affiliation: Institute of Marine Industry / Marine Industry Research Center, Gyeongsang National University
Representative Project No.: 2019-1124; 2020R1A2C10148551361382116530103

Research Goal: To develop network slicing and time-sensitive networking (TSN) technologies for QoS-aware edge handover, real-time communication, and service continuity in autonomous vehicles and wireless edge environments.

Key Contributions: Investigated lightweight slicing over Wi-Fi infrastructures and extended the research toward dynamic TSN and edge handover for autonomous driving, focusing on latency-sensitive networking and robust edge connectivity.

Prototyping Automotive Plug-and-Play Application Environment under Virtualized Resources

Period: Dec 2014 – Dec 2019

Institution: University of Pennsylvania, USA Role: Postdoctoral Researcher Industrial Partner: Toyota Motor Corporation

Developed a plug-and-play (PnP) framework enabling safe and dynamic integration of in-vehicle applications over virtualized computing resources.

Focused on ensuring real-time and safety properties during runtime composition using formal modeling and compositional verification methods, and extended the formal PnP framework to support dynamic resource allocation and system reconfiguration.

Methods for Detecting Interoperability Failures in Dynamically Composed Interoperable Medical Devices

Period: Dec 2014 – (during postdoctoral appointment)

Institution: University of Pennsylvania, USA Role: Postdoctoral Researcher

Developed a formal methodology to generate monitors for integrating medical devices based on the Integrating Clinical Environment (ICE) architecture.

Designed monitoring trees and temporal-logic-based monitoring to ensure that independently developed medical devices can be safely composed to provide comprehensive clinical care.

MBAT – Combined Model-based Analysis & Testing of Embedded Systems

Period: Dec 2012 – Dec 2014

Institution: Aalborg University, Denmark Role: Postdoctoral Researcher Grant: Approx. $220,000 / 3 years

Developed use cases for modeling and analysis of automotive control systems using UPPAAL and UPPAAL SMC.

Proposed a new process algebra, PACoR (Process Algebra of Communicating Resources), and analysis methods based on UPPAAL/UPPAAL SMC, inspired by ACSR. Results were presented at ECRTS 2014 and FMICS 2015, followed by development of a translator from PACoR to UPPAAL for automated analysis.

CRAFTERS – Constraint and Application Driven Framework for Tailoring Embedded Real-time Systems

Period: Dec 2012 – Dec 2014

Institution: Aalborg University, Denmark Role: Postdoctoral Researcher Grant: Approx. $140,000 / 3 years

Developed a methodology for schedulability analysis of various scheduling systems, including hierarchical and probabilistic scheduling systems, using UPPAAL and UPPAAL SMC.

Results were presented in FACS 2013, ERTSS 2014, FACS 2014, TASE 2014, ISORC 2015, and CBSE 2015.

High Risk High Return Project: Formal Modeling and Analysis of Dynamic Production and Consumption of Resources Under Timing Constraints

Period: Jan 2012 – Dec 2012

Institution: KAIST, South Korea Role: Postdoctoral Researcher Grant: Approx. $42,000 / 1 year

Studied various types of resources for cyber-physical systems and developed modeling and analysis methods for resource-constrained systems based on process algebraic approaches. Results were published in REACTION 2012.

Safety Analysis, Verification and Validation for Nuclear Digital Equipment

Period: Jan 2011 – Sep 2011

Company: Korea Nuclear Technology Co., South Korea Role: Verification and Validation (V&V) Engineer Grant: Approx. $25,000 / 6 months

Applied formal methods to verification and validation of requirements and design of a digital hydrogen monitoring system for nuclear power plant protection systems.

FMEA of Youngkwang Nuclear Plant 3,4 Control Systems

Period: Oct 2010 – Nov 2010

Company: VITZROSYS Co., South Korea Role: Verification and Validation (V&V) Engineer

Assisted in reliability analysis of control systems via FMEA, and developed formal requirement and design specifications of control logic for nuclear power plant protection systems using STATEMATE Statecharts.

Formally verified control system models by model checking and validated them through virtual prototyping and simulation.

Study Formal Specification for Assurance of Railway Control Systems

Period: Autumn 2007 – Jun 2009

Institution: Korea Railroad Research Institute, South Korea Role: Verification and Validation (V&V) Engineer Grant: Approx. $34,000 / 2 years

Developed formal requirement and design specifications for components of a railway interlocking system, such as a Distance Control System, using STATEMATE Statecharts, and verified the models via model checking and virtual prototyping.

Formal Method and Testing for Processor Module for Safety-Critical Applications (pCOS)

Period: Jul 2001 – Apr 2008

Institution: Korea Atomic Energy Research Institute, South Korea Role: V&V Engineer and Researcher Grant: Approx. $442,000 / 7 years

Developed formal requirements and design models of a real-time operating system, pCOS (POSCON C/OS), dedicated to a Korean I&C system for nuclear power plant protection systems, using STATEMATE Statecharts.

Formally verified and validated pCOS models through model checking and virtual prototyping, and assisted in preparing certification documents for approval by KINS (Korea Institute of Nuclear Safety).

Audited Verification and Validation for Nuclear Power Plant Control System

Period: Apr 2002 – Jul 2002

Company: U-ri Technology Co., South Korea Role: Verification and Validation (V&V) Engineer Grant: Approx. $17,000 / 4 months

Audited verification and validation documents of distributed control systems (DCS) for Korean I&C nuclear power plant protection systems.

Formal Specification and Verification of Nuclear Power Plant Controlling System

Period: Sep 2000 – Sep 2001

Institution: Korea Electrical Engineering Science Research Institute, South Korea Role: Fundamental Researcher

Studied formal methods for I&C software of nuclear power plant control systems, focusing on rigorous specification and verification techniques.

Development of a Formal Method Tool for Shutdown Control Systems

Period: Feb 1999 – Nov 2000

Company: Han-U Tech Co., South Korea Role: Software Developer Grant: Approx. $17,000 / 1 year

Developed a formal analysis tool for control logic based on the Esterel analysis framework, targeting shutdown control systems in safety-critical domains.