About

About

Inseong (Joseph) Paik

I study Mechanical Engineering at KAIST with a double major in Electrical Engineering. My work focuses on robot dynamics, control, embedded systems, and learning-based robotics, with increasing attention to embodied systems that tie perception, decision-making, and physical interaction into one stack.

Links: GitHub · YouTube

This page is the compact version of what I build and how I work. The blog itself is where I keep the longer notes, implementation details, and project logs.


Current Focus

  • Legged robotics: humanoid locomotion, MPC, contact dynamics, whole-body behavior
  • Robot learning: reinforcement learning, imitation learning, vision-conditioned policies
  • VLA / Physical AI: vision-language-action systems, embodied intelligence, real-world robot behavior
  • Real systems: getting hardware, firmware, control, and software to cooperate under constraints

Education

KAIST
Feb 2025 - Present
B.S. in Mechanical Engineering, Double Major in Electrical Engineering
GPA: 4.10 / 4.30

SASA
Mar 2022 - Jan 2025
Mechatronics, Robotics, and Automation Engineering
GPA: 4.18 / 4.30


Experience

KAIST DRCD Lab

Research Intern · Daejeon, South Korea
Feb 2026 - Jun 2026

  • Researched robot dynamics, convex optimization, and model predictive control for legged robots.
  • Implemented and evaluated control pipelines for humanoid locomotion and dynamic robotic systems.

Selected Projects 🛠️

Convex MPC for Humanoid Robots

Mar 2026 - May 2026 · Research / Personal Project

Built a convex MPC-based humanoid locomotion framework in MuJoCo and C++, with a focus on contact-wrench optimization and single-rigid-body dynamics for stable bipedal walking. I wrote a full technical article covering the mathematics (SRB dynamics, condensed QP, yaw-rotated CoP constraints), solver engineering, and the MPC debugging methodology behind it.

Links: Technical Article · GitHub · Demo Video

Panbot

Jan 2026 - Feb 2026 · Personal Robotics Project

An autonomous pancake-cooking robot built on LeRobot SO-ARM101, combining a staged manipulation pipeline with imitation-learning policies. The system used YOLOv8 segmentation and a ResNet18-GRU state estimator to switch behaviors robustly during pouring, flipping, and plating. I wrote a full technical article covering the perception-triggered runtime, the trigger math and debouncing, the GRU training pipeline, and the ACT policy integration.

Links: Technical Article · Control / System Repo · Vision Repo · Demo Video

mj_ws

Jan 2026 · Personal Software Project

A lightweight MuJoCo RL workspace for rapid robotics experimentation, inspired by IsaacLab-style workflows. I built it to make environment setup, training runs, and logging reproducible with minimal friction.

Links: GitHub · Demo Video · Related Blog Post

Geochemical Exploration Rover Build Project

Sep 2024 - Dec 2024 · Project Lead

Led the end-to-end design and manufacturing of a geochemical exploration rover, including the driving system, control stack, onboard computing, and multi-sensor integration.

Links: Demo Video · Related Blog Post

Articulated Self-Balancing Robot

2025 - 2026 · Club Project

Directed the design and construction of an articulated self-balancing robot, coordinating the mechanical structure, actuator integration, and embedded control implementation.

Acquisition of Fundamental SLAM Materials for Implementing Indoor Navigation Service in School Environments and Establishment of Mobile Robot Driving System

Mar 2023 - Dec 2023 · Personal Project

Built a mobile robot system from scratch using ROS 2, Raspberry Pi, and Arduino for indoor mapping and navigation experiments in school environments.

Links: Demo Video · Related Blog Post

Control-focused system modeling and feedback implementation work using Simulink, Arduino, and real hardware validation.

Links: Demo 1 · Demo 2 · Related Blog Post

EMG-Powered Smart Pillow

Automatic Height Adjustment to Reduce Neck Muscle Fatigue During Sleep

A human-centered mechatronics project using EMG signals to drive automatic pillow height adjustment for improved sleep posture and reduced neck muscle fatigue.

Links: Demo Video


Leadership & Activities 🚀

KAIST MR Robotics Club (2025 - 2026)

Regular Project Team Lead (2025)
Software & Science Team Member, URC 2026

  • In 2025, led a regular club project to design and build an articulated self-balancing robot, coordinating system integration across mechanics, actuation, and embedded control.
  • As a Software & Science team member for URC 2026, contributed to rover development and mission systems for KAIST’s University Rover Challenge team.
  • The team placed 22nd globally in its first year reaching the URC finals, a strong result for a first-year finals run.

SASA R.O.G (Robotics Club)

President (2022 - 2023)

  • Organized foundational robotics education covering Arduino and Raspberry Pi for club members.
  • Ran team-based robotics projects, helping members move from basic embedded practice to complete working systems.
  • Operated robotics booths during SASA public events, presenting projects and engaging visitors through demos and explanations.

Technical Stack ⚙️

AreaTools
ProgrammingC/C++, Python, Swift, CMake, Git
RoboticsROS 2, MuJoCo, Isaac Sim, MATLAB/Simulink
ControlMPC, robot dynamics, convex optimization, feedback control
LearningPyTorch, reinforcement learning, imitation learning, ACT, YOLO
SystemsSLAM, embedded systems, sensor integration, onboard computing

Contact 📬