De Ru Tsai
Robotics Engineer
ROBOTICS EXPERIENCE
Arizona State University, AZ (Aug 2021 - May 2023)
- Evaluated the “Continual Few-Shot Image Classification Model”
(Python, PyTorch, NumPy, OpenCV, Colab)- Implemented classification by building clustering and KNN algorithms under the transformer backbone
- Approached state-of-the-art performance of continual few-shot classification by a fixed classifier and cosine similarity
- Developed the “Robot-to-Robot Charging System”
(Matlab, Mobile Robotics Simulator)- Created the robot-to-robot charging system by implementing the Hungarian matching algorithm
- Saved idle robots by 4 times compared to the traditional charging strategy
- Executed the project “Barrier-Free Area Detection”
(Python, TensorFlow, OpenCV, Colab)- Labeled data, implemented the surface normal filter, and trained the Mask-RCNN model to detect barrier-free area
- Reduced the mask loss by 65% compared to the model without a surface normal filter
- Organized the project “Navigation in Unfamiliar Environments”
(Python, PyTorch, JSON, HDF5, Colab)- Implemented self-adaptive navigation by applying a meta-reinforcement learning model with RGB input
- Improved the success rate by 46% compared to random navigation
- Participated in the project “Grasp Moving Objects in Microgravity”
(Python, PyTorch, MuJoCo, Ubuntu)- Designed the reward functions and robot policies for reinforcement learning in the microgravity environment
- Improved velocity toleration by 4.6 times with discrete samples than with uniform random samples
Industrial Technology Research Institution, Taiwan (Nov 2019 - Jun 2021)
- Launched the “Cyber-Physical Robotic Metal Grinding System”
(C++, OpenCV, Qt, FANUC)- Developed the trajectory compensation plug-in with a laser sensor for an offline trajectory planning software
- Incorporated object detection for recognizing workpieces
- Designed the layout, manipulated the robot arm, and integrated it with equipment, sensors, and PLC
- Developed the “Mold-Free Robotic Roller Hemming System”
(C++, Qt, SolidWorks, FANUC)- Designed the double-roller hemming tool that got rid of multiple molds with the automotive degree adjustment feature
- Planned the hemming trajectories for automotive fabrication
EDUCATION
Arizona State University, AZ (Aug 2021 - May 2023)
- GPA: 4.00 (Graduated with Distinction)
- Degree: Master of Science
- Major: Robotics and Autonomous Systems
(Concentration: Artificial Intelligence) - Courses: Artificial Intelligence / Statistical Machine Learning /
Advances in Robot Learning / Perception in Robotics /
Machine Vision and Pattern Recognition /
Remote Sensing and Synthetic Aperture Imaging /
Multi-Robot Systems / Modeling and Control of Robots /
Linear Algebra / Game Theory: Models and Algorithms
National Chiao Tung University (NCTU), Taiwan (Sep 2016 - Aug 2018)
- GPA: 4.00
- Degree: Master of Science
- Major: Mechanical Engineering
SKILLS
Programming
Python, Matlab, C++, C
Tools
PyTorch, OpenCV, Git, Ubuntu, Arduino, SolidWorks, Google Colab, MuJoCo
Industrial Knowledge
Robotics, Artificial Intelligence, Machine Learning, Computer Vision, Algorithms, Data Structures
OTHER EXPERIENCE
National Fire Agency (Nov 2018 - Oct 2019)
Administrative Assistant, Substitute Military Services
N/MEMS Lab, NCTU (Sep 2017 - Jan 2018)
Teaching Assistant, Mechanical Engineering Laboratory
EXTRACURRICULAR ACTIVITIES
Department Student Association, NCTU (Sep 2013 - Jun 2015)
Manager of Event Planning Department
North Taiwan Area Alumni Association, NCTU (Sep 2013 - Jul 2014)
Manager of Public Relations Department