Kyoungin Baik

I'm an undergraduate in Sungkyunkwan University, majoring in System Management Engineering & Computer Science and Engineering & Mathematics. I'm interested in Reinforcement Learning and Robotics. Currently, I'm working as an intern at RLLAB advised by Prof. Youngwoon Lee.

Email / CV / Github

Research

A significant amount of research is being conducted in robotics. At the end of the day, I believe robots will take the form of humanoids. I am progressing toward that goal step by step.

My current research focuses on the following key approaches:

Autonomous Skill Refinement: Drawing inspiration from the natural self-directed learning observed in humans, this approach is geared toward enabling robots to continually improve their own capabilities. By harnessing the power of parallelized simulation platforms such as Isaac Lab, I am developing methods that allow robots to learn from their own interactions and adapt without relying solely on explicit external feedback. This strategy aims to create systems that are scalable and capable of in-context learning, much like the latest large language models.
Advanced Visual Understanding: Effective robotic perception goes beyond simple data capture—it requires a deep extraction of meaningful visual cues from complex sensory inputs. My work in this area focuses on designing algorithms that learn robust visual representations through diverse simulated scenarios. Utilizing high-fidelity simulation tools, I seek to empower robots with a vision system that generalizes well across varied environments, facilitating better interpretation and interaction with the world.
Dynamic In-Field Adaptation: Real-world environments present unpredictable challenges such as partial observability and shifting conditions. To bridge the gap between controlled simulations and real-life applications, I am developing adaptation techniques that enable robots to adjust their behavior on the fly. By integrating real-time feedback with in-context learning strategies, this approach ensures that robotic systems remain robust and effective even when faced with unforeseen complexities.

By integrating these innovative directions, my research strives to bridge the gap between computational intelligence and cognitive adaptability, paving the way for robotic systems that are not only scalable and generalized but also capable of continuous, self-driven evolution.

Research Experinces and Projects

	Mathematical Optimization theory research Undergraduate Research Intern at Sungkyunkwan University, Mathematics This project explored diverse optimization algorithms, including Stein Variational Gradient Descent (SVGD) and Model Predictive Path Integral (MPPI). I conducted a comprehensive literature review and implemented these algorithms in Python. I also developed a new algorithm that combines SVGD and interior point method to improve the convergence rate and stability of SVGD.
	Max Planck Institute for Demographic Research (MPIDR) Korea-Germany Junior Research Fellowship Program 2023 Visiting Researcher as Data Scientist & Machine learning Engineer at MPIDR, Rostock, Germany. I worked on the project regarding twitter data by using LLM with Dr. Jisu Kim. I developed a new framework for analyzing the demographic data of Twitter users. I also implemented a new algorithm that combines LLM and clustering methods to improve the performance of LLM.
	Metaverse Competition Ministry of Education and National Research Foundation of Korea, COSS I participated in the Metaverse Competition held by the Ministry of Education and National Research Foundation of Korea. I developed a metaverse platform that enables users to interact with each other by using proxy server and python signal to connect between Roblox and Discord.

Website template from Jon Barron.