JEONJU · KOREA · EST. 2026
Physical
AI Lab
Bridging machine intelligence and the physical world — embedded systems, autonomous robotics, and real-time AI at the edge.
AFFILIATED INSTITUTIONS
Research Areas
01
⚡Edge AI & Deployment
TensorRT, ONNX, quantization — making learned models fast and lean enough to run on constrained hardware at the edge.
02
📡Sensor Fusion & Perception
Kalman filters, multi-modal sensing, depth cameras, and vision pipelines that let machines reliably perceive the real world.
03
🤝Human-Robot Interaction
Designing safe, intuitive interfaces between people and robotic systems — from gesture recognition to natural language commands.
04
🛠️AI-Assisted Development
Tools and methodologies for integrating AI into the software development process, enhancing productivity and code quality.
05
📚Physical AI Education
Curriculum design, competency frameworks, and pedagogy for teaching Physical AI at the undergraduate and graduate level.
Research
A Low-Cost Sentinel-1 SAR and Google Earth Engine Pipeline for Flood Detection in Mid-Size Cities: Validation on the July 2023 Jeonju Flood Event
중소 도시 홍수 탐지를 위한 Sentinel-1 SAR 및 Google Earth Engine 기반 저비용 파이프라인: 전주시 2023년 홍수
AI-Enhanced Cell-Free CRISPR Diagnostics for Decentralized Biosensing
AI 기반 무세포 CRISPR 진단법을 활용한 분산형 바이오센싱
Design and Implementation of a Low-Cost IoT-Based Smart Irrigation System with Real-Time Monitoring
실시간 모니터링 기능을 갖춘 저비용 IoT 기반 스마트 관개 시스템의 설계 및 구현
Design and Performance Characterization of a Low-Cost Rough Vacuum System for Hands-On Semiconductor Education
실습 중심의 반도체 교육을 위한 저비용 저진공 시스템 설계 및 성능 특성 분석
Machine Learning Prediction of Melting and Boiling Points of Elements Using Atomic Descriptors
원자 설명자를 이용한 원소의 녹는점 및 끓는점 예측을 위한 기계 학습
A Conceptual Framework for Ethical Digital Consent and Data Privacy in Web Services
웹 서비스에서 윤리적 디지털 동의 및 데이터 프라이버시를 위한 개념적 프레임워크
Projects
A Context-Aware Korean Slang and Idiom Dataset
A dataset of 189 Korean slang phrases with appropriate English translations was created and labeled with sentiment, inte...
Normalized Handwritten Hangul Graphemes Dataset (NHHG) + Pangrams
89,100 handwritten Hangul graphemes (2,200 per consonant, 1,100 per vowel grapheme) were collected from 110 freshmen uni...
KSPAI — Korean Society for Physical AI (founding)
Founding member and organiser of the Korean Society for Physical AI (한국피지컬AI학회). The society promotes research exchange,...
Manchu Ancient Script OCR Dataset (MASD)
[Our paper](https://doi.org/10.56977/jicce.2024.22.1.80) demonstrated that it is possible to semi-manually gather ancien...
Handwritten Cherokee Script OCR Dataset (CSD)
[Our paper](https://doi.org/10.56977/jicce.2024.22.1.80) demonstrated that it is possible to semi-manually gather ancien...
Curriculum
Four structured tracks from first principles to advanced robotics. Fully bilingual EN / 한국어.
Doing Academic Research
Learn how to do all kinds of academic research: papers, projects, capstone design, conference presentations, journal publications.
12 units EN · 한국어Foundations of Physical AI
Sensors, actuators, and the full AI stack from first principles. No prior ML experience required.
8 units EN · 한국어Edge AI and Embedded Systems
TinyML, RTOS, and real-time inference on constrained hardware. Covers Arduino, ESP32, and Raspberry Pi.
10 units EN · 한국어Robotic Systems and Control
Kinematics, ROS 2, and perception–action loops. Build and program a 4-DOF robot arm from scratch.
12 units EN · 한국어Datasets and Benchmarks
Lab-produced open datasets for Physical AI research. All released under CC-BY 4.0.
3 units EN · 한국어Arduino: Foundations of Physical Computing
Learn physical computing with Arduino — from blinking LEDs to sensor integration and motor control. No prior hardware experience required.
8 units EN · 한국어ESP32: IoT and Wireless Physical AI
Build Wi-Fi-enabled sensors, MQTT pipelines, and edge AI applications using the ESP32 microcontroller.
8 units EN · 한국어Jetson Nano: GPU-Accelerated Physical AI
From JetPack setup to real-time neural inference — run YOLO, segmentation models, and ROS 2 nodes on NVIDIA Jetson Nano.
8 unitsCourse plug-in modules
Physical AI add-ons for courses not primarily focused on AI — drop into any syllabus in 1–3 weeks.
AI-aware memory management
Efficient C++ patterns for inference workloads — stack vs. heap, arena allocators, and avoiding dynamic memory on bare-metal targets.
3 weeksSensor fusion fundamentals
Kalman filters and multi-sensor pipelines — plug into any IoT course to add a Physical AI dimension.
2 weeksTime-series data for physical systems
InfluxDB and real-time sensor logging — adds a physical AI data layer to any database course.
2 weeksNeural network hardware mappings
How circuit theory underlies Physical AI acceleration chips — MAC arrays, memory bandwidth, and power analysis.
1 weekNotes
RESEARCH OPPORTUNITIES
Work with the lab
I work with motivated undergraduates on conference papers, open datasets, and Physical AI projects — Arduino robotics, edge ML, embedded systems. No prior AI background required. If you have a capstone project with potential, or just want to learn by building something real, reach out.
About
PRINCIPAL INVESTIGATOR
The Physical AI Lab is a research group based in Jeonju, Korea, working at the intersection of machine learning, embedded hardware, and physical systems.
I teach across five Korean universities and run an open research practice — papers, datasets, and code are published wherever possible. I'm also a founding organiser of KSPAI (한국피지컬AI학회), the Korean Society for Physical AI.
Wyoming native. Laid back. Will talk robotics over coffee.
AFFILIATED INSTITUTIONS