I am a researcher passionate about developing intelligent robots with the ability to plan efficiently, learn from past experiences, and reason about their decisions and interactions with other agents. In particular, I design algorithms for task and motion planning, as well as multi-robot systems, aiming for solutions that are both theoretically sound and practically valuable.
I am a postdoctoral fellow in the Learning Agents Research Group within the Computer Science Department at the University of Texas at Austin, hosted by Prof. Peter Stone. Previously, I was a postdoctoral associate at MIT CSAIL, hosted by Prof. Tomás Lozano-Pérez and Prof. Leslie Pack Kaelbling.
I received my Ph.D. degree from Virginia Tech under Prof. Pratap Tokekar in 2019. I received my M.S. and B.S. degrees from Korea University in 2013 and 2011, respectively.
My favorite quote from David Blackwell:
"Basically, I'm not interested in doing research and I never have been... I'm interested in understanding, which is quite a different thing. And often to understand something you have to work it out yourself because no one else has done it."
Research
My current research is primarily motivated by addressing long-horizon robot decision-making problems, such as cooking a meal or cleaning a dirty room. Such long-horizon tasks present challenges in selecting high-level actions, like picking up a cup, and in effectively planning how to achieve these actions while satisfying physical and geometric constraints. Existing algorithms are prohibitively expensive to scale up beyond the smallest of problems. To tackle this challenge, I design general-purpose task and motion planning algorithms with theoretical guarantees and improving their efficiency by (1) leveraging metareasoning methods, (2) investigating long-horizon dependencies with the help of machine learning techniques, and (3) distributing tasks among multiple robots.
Asynchronous task plan refinement for multi-robot task and motion planning
IEEE International Conference on Robotics and Automation (ICRA 2024)
A new multi-robot task and motion planning formulation that eliminates the need for pre-discretization of the space and synchronous actions among robots.
Motion planning (in)feasibility detection using a prior roadmap via path and cut search
Robotics: Science and Systems (RSS 2023)
Iterative pathfinding and cut finding effectively reduce the search space, making the determination of whether a prior roadmap contains a feasible solution more efficient.
A survey of decision-theoretic approaches for robotic environmental monitoring
Foundations and Trends in Robotics 2023
A comprehensive survey on robotic environmental monitoring, covering environmental representations and properties, tasks, and decision-theoretic tools.
Learning when to quit: meta-reasoning for motion planning
IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS 2021) Finalist for Best Cognitive Robotics Paper Award
Being able to trade-off between plan quality and computation time lets you build a more efficient planner.
Multi-resolution POMDP planning for multi-object search in 3D
IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS 2021) Winner of Best Robocup Paper Award
The challenge of intractable partially observable 3D object search can be alleviated through a multi-resolution belief space representation.
Reactive task and motion planning under temporal logic specifications
IEEE International Conference on Robotics and Automation (ICRA 2021)
When robots work with humans, linear temporal logic allows the modeling of human interventions, while a behavior tree enables the robot to execute reactively.
Environmental hotspot identification in limited time with a UAV equipped with a downward-facing camera
IEEE International Conference on Robotics and Automation (ICRA 2021)
Unlike standard multi-armed bandit settings, the arms' rewards are correlated, and the reward amount depends on the altitude of a UAV.
GM-PHD filter for searching and tracking an unknown number of targets with a mobile sensor with limited FOV
IEEE Transactions on Automation Science and Engineering (T-ASE 2021)
IEEE International Conference on Robotics and Automation (ICRA 2017)
The explicit boundary of limited sensing range is incorporated into the GM-PHD filter, which estimates both the number and spatial density of targets.
Game tree search for minimizing detectability and maximizing visibility
Autonomous Robots (AURO 2021)
IEEE International Conference on Robotics and Automation (ICRA 2019)
Minimax search tree and Monte-Carlo search tree methods for discrete, sequential, two-player, zero-sum games that involve trading off conflicting objectives.
Distributed assignment with limited communication for multi-robot multi-target tracking
Autonomous Robots (AURO 2020)
IEEE International Conference on Robotics and Automation (ICRA 2018)
The reduction of multi-robot multi-target tracking problems to mixed packing and covering problems enables the use of local algorithms for distributed problem-solving.
A competitive algorithm for online multi-robot exploration of a translating plume
IEEE International Conference on Robotics and Automation (ICRA 2019)
A recursive depth-first-search algorithm for exploring the unknown shape of a moving plume, with provable competitive ratio guarantees.
Team VALOR’s ESCHER: A novel electromechanical biped for the DARPA Robotics Challenge
Journal of Field Robotics (JFR 2017)
Virginia Tech's electric series compliant humanoid participated in the finals of the DARPA Robotics Challenge.
Hierarchical sample-based joint probabilistic data association filter for following human legs using a mobile robot in a cluttered environment
IEEE Transactions on Human-Machine Systems (T-HMS 2016)
Correlations among targets are explicitly modeled for target tracking, a consideration that is generally treated independently in the tracking literature.