index.html

---
layout: base
title: Home
---

<h1>Home</h1>

<p>My name is <strong>Solgyu Lee</strong>, a junior student at Soongsil University in Seoul, South Korea.</p>

<p>My research interests lie in the intersection of <strong>Robotics</strong> and <strong>Multi-Agent Reinforcement Learning (MARL)</strong>. I am currently working on <a href="/study/marl">multi-agent systems</a>, <a href="/study/rl">reinforcement learning</a>, and <a href="/study/robotics">robotics applications</a>.</p>

<p>Meanwhile, I am a passionate learner who enjoys building robots and exploring new technologies. Check out my <a href="/projects">projects</a> to see some of my work, or visit my <a href="/journey/all">journey</a> to follow my learning adventures.</p>

<p><strong>Contact:</strong> <a href="mailto:leesg0107@gmail.com">leesg0107@gmail.com</a> | <a href="https://github.com/leesg0107">GitHub</a> | <a href="https://linkedin.com/in/leesg17">LinkedIn</a></p>

<hr>

<h2>Research Focus: Heterogeneous Multi-Robot Systems</h2>

<p>My primary interest lies in <strong>Multi-Agent Reinforcement Learning (MARL)</strong> and its applications in <strong>heterogeneous robot systems</strong>. I believe the truly transformative moment in robotics will come when we can effectively replace labor-intensive human work with intelligent robotic systems.</p>

<p>Consider agriculture, for example: planting seeds across vast fields, monitoring crop health, and harvesting—all require different capabilities. <strong>No single type of robot can handle all these complex tasks</strong>. What we need are diverse robots with complementary skills working in coordination.</p>

<p>This is why heterogeneous multi-robot systems fascinate me. Building such systems requires integrating multiple technologies beyond just basic RL or MARL algorithms:</p>

<ul>
  <li><strong>Graph Neural Networks (GNN)</strong> for modeling agent relationships</li>
  <li><strong>Large Language Models (LLM)</strong> integration for high-level task understanding</li>
  <li><strong>Hierarchical architectures</strong> for multi-level coordination</li>
  <li><strong>Hardware integration</strong> bridging simulation and reality</li>
</ul>

<p>I'm actively studying and researching these interconnected areas, going beyond single-algorithm approaches to build truly practical systems.</p>

<h2>Skills</h2>

<table class="skills-table">
  <thead>
    <tr>
      <th>Programming</th>
      <th>Frameworks & Tools</th>
      <th>Hardware</th>
    </tr>
  </thead>
  <tbody>
    <tr>
      <td>Python</td>
      <td>PyTorch</td>
      <td>Soldering</td>
    </tr>
    <tr>
      <td>Rust</td>
      <td>ROS2</td>
      <td>Robot Design</td>
    </tr>
    <tr>
      <td>C++</td>
      <td>Isaac Sim</td>
      <td></td>
    </tr>
    <tr>
      <td></td>
      <td></td>
      <td></td>
    </tr>
  </tbody>
</table>

<h2>Why I Aim to Be a Generalist</h2>

<p>What drew me to reinforcement learning was its elegance: <strong>learning optimal behaviors through trial and error</strong>. But through countless simulation experiments, I realized something crucial—<strong>simulations alone aren't enough</strong>. To make robotics meaningful, I needed to build real robots. This meant diving into soldering, electrical engineering, mechanical design. The more challenging it gets, the more I'm drawn to it.</p>

<p>This journey taught me that <strong>technologies are tools for solving problems, not ends in themselves</strong>. Reinforcement learning, like any machine learning technique, is just one approach—and it too can always be replaced by better ideas. What matters isn't mastering a specific algorithm—it's understanding the problem deeply: What's been tried? What failed? How can we do better? Rather than falling into "tutorial hell," I've embraced <strong>learning through building and failing</strong>.</p>

<p>My goal is to become a <strong>generalist</strong>—someone who can traverse disciplines, connect ideas across fields, and tackle problems from first principles without being constrained by narrow specialization.</p>

<h2>Biography</h2>

<div class="biography">
  <strong>2002.1.7</strong> Born in Seoul, Korea 🇰🇷<br>
  <strong>2021.3</strong> Entered Soongsil University<br>
  <strong>2022.5.2</strong> Military service begins<br>
  <strong>2023.10.27</strong> Discharged from military service<br>
  <strong>2024.3</strong> Returned to university as sophomore<br>
  <strong>2025</strong> Currently junior, focusing on MARL & Robotics / President of Soongsil American Football team CRUSADERS
</div>

<h2>Who Am I</h2>

<p>I'm energetic, a problem-solver, and someone who genuinely enjoys the work—whether it's studying algorithms, building robots, or diving into new fields. Beyond the lab, I play for Soongsil University's American football team and love staying active through weightlifting, running, and basketball.</p>

<p>My greatest strength? <strong>I don't shy away from anything</strong>. Every experience teaches me something valuable, even the difficult ones. Why not try it? (As long as it's not completely foolish.) This mindset lets me jump into unfamiliar situations without hesitation.</p>

<p>My weakness? Being a bit of a "yes man"—sometimes I take on more than I should without fully thinking through the consequences. I'm actively working on this through hard-learned lessons, finding the balance between saying yes to opportunities and being strategic about commitments.</p>

<p><strong>Life Philosophy:</strong> <em>Every challenge is an opportunity to grow. I've learned that embracing both successes and setbacks with curiosity rather than frustration leads to deeper understanding and better solutions.</em></p>