cv

PhD Candidate

Georgia Institute of Technology

Robotics and Mechanical Engineering

• Advisor: Dr. Patricio A. Vela
• Co-Advisor: Dr. Jun Ueda

Master

University of California, Berkeley

Mechanical Engineering

Bachelor

Chongqing University

Mechanical Engineering

Teaching Practicum

Georgia Institute of Technology, ME

ME 3017: System Dynamics. Supervised by Dr. Jun Ueda

• Participated in the creation of course materials, assignments, and exams while also offering guidance through office hours. Delivered portions of lectures and supported undergraduate students in addressing their academic inquiries.

ORS and VIP Undergraduate Research Mentor

Georgia Institute of Technology, ECE

Supervised and organized 8+ undergraduate research projects on vision-based navigation with SLAM, perception, planning, teleoperation, control, deep learning, software, hardware design, etc. Offering help and support to undergraduate researchers.

Perception Engineer Intern

Seres (SF Motors), ADAS Team

Supervised by Chongyu Wang and Fan Wang

• Implemented C++ OpenCV algorithm to achieve stop-line and traffic light detection through classical computer vision.
• Contributed to deep learning traffic detection and data preparation.
• Deployed classical and learning-based algorithms in autonomous driving field tests.
• Assisted in completing camera installation, sensor calibration, and real-time image acquisition.

Graduate Research Assistant

Georgia Institute of Technology, IVALab

Hierarchical Stereo Navigation with Sparse Representation. Supervised by Dr. Patricio A. Vela

• Created a sparse ego-centric perception space from stereo cameras to describe local environments and track temporal sensing information for real-time motion planning and collision checking that has five times faster computational efficiency and scalability among workstations and lightweight embedded devices.
• Designed safety-guaranteed motion planning methods to achieve 0% collision rates for holonomic and nonholonomic dynamics involving model predictive control, potential fields, and control barrier functions, which improves safety performance of classical planning techniques in configuration space.
• Established an image-based trajectory tracking method with VSLAM to improve trajectory tracking accuracy by 28%.
• Implemented a vision-based navigation framework (GPF-BG) for quadrupedal robots to obtain 10% more success rates.
• Developed real-time navigation system architectures containing perception, planning, SLAM localization, and control modules for different platforms: mobile robots, mobile manipulation robots, snake-like robots, and quadrupedal robots.
• Conducted quantitative research on navigation performance in ROS/Gazebo simulation and real robots (Turtlebot, LoCoBot, Stretch, Unitree A1) with stereo cameras, depth cameras, laser scanners, and LiDAR.
• Trained deep learning neural networks to intelligibly select ego-centric collision-free trajectories from stereo images, which involves machine learning and computer vision.
• Research to deploy reinforcement learning models for navigation and object searching in a mobile manipulation task.

Graduate Researcher

University of California, Berkeley, MPC Lab

Fault Tolerant Control in Autonomous Driving, Perception. Advised by Dr. Francesco Borrelli

• Developed the main sensor data association algorithm in Python with an external optimization solver.
• Tested the sensor association algorithm in simulation and on a real autonomous driving car.

Graduate Teaching Assistant

Georgia Institute of Technology, ME

ME2110: Creative Decisions and Design. Supervised by Dr. Thomas Kurfess and Dr. Christopher Saldana

• Worked in collaboration to create course materials, assignments, and exams while offering office hours.
• Instructed on mechatronics and machining training while overseeing machining open labs.
• Acted as the lead TA for a semester, organizing TA training and open labs, and coordinating the final competition.