JINWOOK HUH

I am currently a Senior Machine Learning Engineer at Motional. Throughout my career, I have focused on the research and development of machine learning solutions for robot manipulation and ML-based planners for autonomous vehicles. I earned my PhD from the GRASP Lab at the University of Pennsylvania under the mentorship of Prof. Daniel Lee. Prior to that, I completed my BS and MS at POSTECH in Korea.

My research interests include artificial intelligence, machine learning, motion planning, robotic manipulation, and autonomous vehicle systems. I also served as a Staff Researcher (Tech Lead) at Samsung AI Center in New York, where I led and executed home robotics projects. Additionally, I have research experience with military unmanned vehicle systems at the Agency for Defense Development in Korea.

For more details about my qualifications and experience, please refer to my CV.

The c2g-HOF architecture consists of a cost-to-go function over the configuration space represented as a neural network (c2g-network) as well as a Higher Order Function (HOF) network which outputs the weights of the c2g-network for a given input workspace.

This approach is useful for continuous trajectory generation of a high dimensional robot manipulator and also non-holonomic systems directly from the workspace description. It can generate a trajectory, avoiding obstacles in cluttered environments, within 0.1 seconds.

dynamic-obstacles.mp4

isaacgym.mp4

RAMP: Hierarchical Reactive Motion Planning for Manipulation Tasks Using Implicit Signed Distance Functions

Vasileios Vasilopoulos, Suveer Garg, Pedro Piacenza, Jinwook Huh and Volkan Isler,, IEEE/RSJ International Conference on Robotics and Systems (IROS), October, 2023. Project page

Real-time Simultaneous Multi-Object 3D Shape Reconstruction, 6DoF Pose Estimation and Dense Grasp Prediction

Shubham Agrawal, Nikhil Chavan-Dafle, Isaac Kasahara, Selim Engin, Jinwook Huh, Volkan Isler, IEEE/RSJ International Conference on Robotics and Systems (IROS), October, 2023.

Pick2Place: Task-aware 6DoF Grasp Estimation via Object-Centric Perspective Affordance

Zhanpeng He, Nikhil Chavan-Dafle, Jinwook Huh, Shuran Song, Volkan Isler, IEEE International Conference on Robotics and Automation (ICRA), 2023

Cost-to-Go Function Generating Networks for High Dimensional Motion Planning

Jinwook Huh, Volkan Isler, and Daniel Lee, Proc. 2021 IEEE International Conference on Robotics and Automation (ICRA), pp. 8480-8486, June 2021.

Learning to Generate Cost-to-Go Functions for Efficient Motion Planning

Jinwook Huh, Galen Xing, Ziyun Wang, Volkan Isler, and Daniel Lee, “Learning to Generate Cost-to-Go Functions for Efficient Motion Planning,” Proc. 2020 International Symposium on Experimental Research (ISER), pp. 555-565, 2020.

ICRA21__home_papercept_ras.papercept.net_www_conferences_conferences_ICRA21_submissions_1434_VI_fi.mp4

ISER20_0016_VD_fi.mp4

Learning Continuous Cost-to-Go Functions for Nonholonomic Systems

Jinwook Huh, Volkan Isler, and Daniel Lee, Proc. 2021 IEEE/RSJ International Conference on Robotics and Systems (IROS), October, 2021.

IROS21_1733_VI_fi.mp4

Probabilistically Safe Corridors to Guide Sampling-Based Motion Planning

Jinwook Huh, Omur Arslan, and Daniel Lee, “,” Proc. 2019 International Symposium on Robotics Research (ISRR), October. 2019.

video_SGRRT_mm2.mp4

Efficient Sampling with Q-Learning to Guide Rapidly-Exploring Random Trees

Jinwook Huh and Daniel Lee, IEEE Robotics and Automation Letters, Vol 3, Issue 4, pp 3868-3875, October, 2018.

icra_2018_qs_rrt_final.mp4

Constrained Sampling-Based Planning for Grasping and Manipulation

Jinwook Huh, Bhoram Lee, and Daniel Lee, Proc. of 2018 IEEE International Conference on Robotics and Automation (ICRA), pp. 223-230, May 2018

How fast can you assemble and disassemble the robot?

assembling_small.mp4

Google Sites

Report abuse