Biography
I am a Ph.D. candidate at UC Berkeley under the supervision of Prof. Masayoshi Tomizuka. My research interest lies in computer vision, motion planning and control with the focus on robotic manipulation and human-robot collaboration. I am recently interested in perception-driven motion planning and control with the application to task-oriented manipulation and in-hand manipulation.
Interests
- Computer Vision
- Motion Planning
- Control
- Machine Learning
- Robotic Manipulation
Education
-
Ph.D. Candidate in Mechanical Engineering, 2021 - now
University of California, Berkeley
-
M.S. in Mechanical Engineering, 2017 - 2019
National Taiwan University
-
B.S. in Mechanical Engineering, 2013 - 2017
National Taiwan University
Publications
Experience
Research Assistant
- Demonstrated online few-shot learning in 2020 Taiwan Innotech Expo where a robotic arm firstly picks up an unseen object in front of a camera and an object detector can be improved according to the new captured image
- Built a novel end-to-end 6-DoF grasp detector that is twenty times faster and more accurate than previous end-to-end learning-based approaches on unseen cluttered objects
- Achieved robotic assembly tasks by developing an algorithm that can detect 6-DoF grasps and planned a series of actions, such as grasping and re-grasping, based on the subsequent task
Graduate Researcher
- Thesis Topic: Robot Grasping in Clutter using Instance Segmentation and Representation Learning
- Proposed a novel learning-based approach that can detect precise and accurate grasps on dense clutter scene based on noisy RGB-D image and outperformed previous end-to-end learning-based approaches on bin-picking tasks
- Led a team of 3 people to implement the algorithm on a system consisted of the robot arm with a two-jaw parallel gripper and Kinect V2
- Collaborated with LNC Technology, an automation company in Taiwan, for demonstrating random bin picking at 2019 Taiwan Automation Intelligence and Robot Show
Intern
- Designed an image processing algorithm that can detect black and white stones on the chessboard
- Helped robot play five-in-a-row with human by adopting minimax algorithm and alpha-beta pruning
Undergraduate Researcher
- Led a team of 4 people to model the dung-rolling behavior of dung beetle and developed corresponding dung beetle robot that can reliably roll a ball
- Designed trajectories and gait of each leg by observing how dung beetles synchronize their legs during the dung-rolling motion
- Optimized the execution time of inchworm robot for wall-climbing and enhanced the speed of the movement by 60%
Intern
- Developed a pick-and-place system that can help robot arm place chips onto different plates after AOI machine defined the chips as either normal or defective ones
- Designed a system consisted of robotic arm and CCD that can detect defects on laptop
Awards
Gold Medal (First Place) in HIWIN Master’s Thesis Award
Presented research of robot grasping, ranked 1st out of 100 researches and awarded 35K USD
Gold Medal (First Place) in SPINTECH Technology Thesis Award
Presented research of dung beetle robot and ranked 1st out of 22 teams
Best Presentation Paper Award
Presented research of dung beetle robot and received compliments from the committee members