Giseop Kim
paulgkim@kaist.ac.kr
NOTE
About Me
2nd year Ph.D. student, advised by Prof. Ayoung Kim.
IRAP Lab, Civil and Environmental Dept., KAIST, South Korea.
Mission
Towards great robotic autonomy in the wild.
Research Interests
Large-scale, Long-term, Robust LiDAR-based Robot Mapping in Complex Urban Sites.
Spatio-temporal and Semantic Understanding of Place.
Publications
Selected
2022 (plan, the final year)
2021 (plan)
tldr: Long-term LiDAR Mapping in Changing Outdoors
Giseop Kim, Younghun Cho, and Ayoung Kim, LT-mapper: An Open Framework for Robust Long-term Multi-session LiDAR Map Construction, journal in preparation. [Paper (TBA), Code (TBA), Video (TBA) / Datasets (TBA): KAIST ParkingLot dataset, Sejong 1-year dataset]
including Long-term LiDAR Map Management
Removert v2: Long-term Static Point cloud Map Construction and Multiple Level of Change Detection, conference in preparation. [Paper (TBA), Code (TBA), Video (TBA) / Datasets (TBA): Urban 3D Object Ephemerality dataset]
tldr: Improved LiDAR Place Recognition
Giseop Kim, Sunwook Choi, and Ayoung Kim, Scan Context++: Rotation and Lateral Invariant Spatial Descriptor for Robust and Fast 3D Place Recognition, journal in preparation. [Paper (TBA), Code (TBA), SLAM code (TBA, SC-LeGO-LOAM v2)] / joint work with NAVER LABS
2020
tldr: LiDAR Static Map Construction
tldr: LiDAR + Radar Dataset in Complex Urban Sites
Giseop Kim, Yeong Sang Park, Younghun Cho, Jinyong Jeong, and Ayoung Kim, MulRan: Multimodal Range Dataset for Urban Place Recognition, IEEE International Conference on Robotics and Automation (ICRA), 2020. [Dataset site, Paper, Slide, Video, Radar Place Recognition Code (TBA)]
2019
tldr: Long-term LiDAR Localization in Changing Outdoors
tldr: Urban Site Analysis using LiDAR
Giseop Kim, Ayoung Kim, and Youngchul Kim, A new 3D space syntax metric based on 3D isovist capture in urban space using remote sensing technology, Computer, Environment and Urban Systems, 2019. [Online] / joint work with urban design lab in CEE
2018
Others
2021 (plan)
tldr: Radar SLAM
Yeong Sang Park, Giseop Kim, ..., and Ayoung Kim, Radar SLAM ..., journal in preparation. [Paper (TBA), Code (TBA)]
2020
tldr: Learning-based LiDAR odometry
tldr: Camera-sonar style transfer-based multimodal matching
Hyesu Jang, Yeongjun Lee, Giseop Kim, Ayoung Kim, CNN-based Opti-Acoustic Transformation for Underwater Feature Matching, Special Issue: Underwater Robotics, Journal of Korea Robotics Society, 2020. [Online]
2019
tldr: a preliminary report of the ICRA2020 paper
Younggun Cho, Giseop Kim, and Ayoung Kim, DeepLO: Geometry-Aware Deep LiDAR Odometry, arXiv, 2019. [Online (arxiv), Video] ➡ moved to the ICRA 20 paper.
tldr: a preliminary report of the JKRoS2020 paper
Hyesu Jang, Giseop Kim, Yeongjun Lee and Ayoung Kim, CNN-based Approach for Opti-Acoustic Reciprocal Feature Matching. In ICRA Workshop on Underwater Robotics Perception, Montreal, May, 2019. [Project page, Online]
2018
tldr: a preliminary report of the RAL2019 paper
Giseop Kim, Byungjae Park and Ayoung Kim, Learning Scan Context toward Long-term LiDAR Localization, In ICRA Workshop on Long-term Autonomy and Deployment of Intelligent Robots in the Real-world, Brisbane, May. 2018. (Best paper award) [Download, Online]
2017
tldr: a preliminary report of the CEUS2019 paper
Giseop Kim, Hyunchul Roh, Youngchul Kim, Ayoung Kim, Sensor Data-driven Urban Site Analysis using Point Cloud from Urban Mapping System, ICRA Late Breaking Results (poster only, non-publishing), 2017. [Project page]
Dissertation
Giseop Kim, Range sensor-based Long-term SLAM for a Mobile Robot in Complex Urban Sites (tmp), Ph.D thesis, KAIST, 2023 (expected). [Download (TBA)]
Giseop Kim, Isovist-induced Robust LiDAR Localization, MS thesis, KAIST, 2019. [Download]
Personal Projects
Done
Scan Context: A real-time LiDAR place recognition on CPU (code of the IROS2018 and the journal2021)
Summary: A fast and robust LiDAR global localizer as well as loop detector
Codes (v2, TBA) [C++, Matlab]
SC-LeGO-LOAM: A light-weight LiDAR SLAM system (C++)
Summary: LeGO-LOAM (odometry) + Scan Context (loop detection) + GTSAM (robust back-end)
PyICP SLAM: Python-only ICP-based SLAM
Undergoing
(By 2020-09-30) Removert (code of the IROS2020)
Summary: using scans and SLAM poses, makes static map, removing dynamic points, parse dynamic object points.
Codes (TBA) [C++, Matlab]
(By 2020-12-31) LT-mapper (code of the journal2021, in preparation)
Summary: An open framework for LiDAR-based Lifelong mapping
Codes (TBA)
(By 2020-12-31) Real-time Radar SLAM (code of the ICRA2020 and the journal2021)
Summary: PhaRaO (odometry) + Radar Scan Context (loop detection)
Codes (TBA)
Projects in charge (selected)
2020.01 - 2021.12: Bilateral Research Exchange Support Program in 2020 (Czech Republic), from NRF, South Korea
Project Title: Towards Long-term Autonomy through Introduction of the Temporal Domain into Spatial Representations used in Robotics
Summary: A joint research about robot long-term autonomy with Prof. Tomas Krajnik, CTU.
2018.05 - 2019.12: Project from MOLIT, South Korea
Project Title: Autonomous urban analysis technology by using 3D sensor data
Summary: A research on development of sensor data-driven urban visibility measure and automation
2018.06 - 12: Venture Research Program for Master’s and PhD Students in the College of Engineering, KAIST, South Korea
Project Title: 3A Localization: Anytime, Anywhere, and Anyview Robot Localization at Complex Urban
Summary: A long-term and large-scale LiDAR localization method
Tools
C/C++, Matlab, Python / Pytorch, Tensorflow / Ubuntu, ROS, Git, SVN, Docker, LaTeX, ...
Korean (native), English (fluent)
Awards
2019 ICRA 2019 RAS Travel Grant
2018 Best paper award at ICRA 2018 LTA workshop
Services
Reviewer
IROS 2020, UR 2020, ICRA 2020 (delegated), IROS 2019 (delegated)
Community
2019.11 - : Facebook group SLAM Korea manager and contents provider
Teaching Assistant
2020 Spring: CE481 (Sensor-based spatial intelligence a.k.a SLAM 101 - syllabus), CEE, KAIST
2016 Spring : CE352 (Signal and System for Construction IT), CEE, KAIST
Talks
updated Jul 10. 2020