Work done in Institute of Information Science - Academia Sinica
I used LabVIEW to write an algorithm letting Dr.Robot avoid obstacle and Enhancing user interface for more friendly operating system.
Latest Projects
Works done in Centrum Embedded Systems
I worked in Centrum Embedded Systems Corporation (CES) which engages mainly in the designing and manufacturing of industrial mobile computers and DJ Mixer. As an engineer, I used software “Blender” and “LibreCAD” to model the object to a very realistic level, simulate its physics action, and do the animation to refine my design. I also used Python to program microchips.
PROJECTS
Below are some projects that I've done before. Feel free to take a look!
I participated in the “2014 American Society of Mechanical Engineers Student Professional Development Conference”, in which my team built our own helicopter that could load up to two kilograms from design. We designed it using Solidworks and wrote the program into Arduino.(Although the report is written in Chinese, it still reveals our work from modeling to programing!)
UAV video (Crossing a 1.5m height line which couldn't be seen in the video)
Robockey Contest
Fast and strong game!(left)
Be able to turn back to get the puck and score again!
Macine Perception Project at Upenn
I used MATLAB to do edge detection of checker board and minimize Geometric error and then reconstruct the camera poses.
Robockey Project at Upenn (Second Price)
• Used C with M2 microcontroller to do path planning, trajectory, wireless communication.
• Used MATLAB to do robot’s localization simulation and turned it into C.
• Designed and built driver circuit.
Advanced Robotics
Quadrarotor simulation
Advanced Robotics: Quadrarotor Project at Upenn
Implement high level controller with numerous PID, then simulated environment and did trajectories and path planning.
Automatic Face Replacement in video in Computer Vision
• Implemented face keypoints detection and matching using shape context and Pictorial Structure.
• Performed face warping using TPS and poisson blending to replace the face.
Emotional Prediction Using Data from Twitter in Machine Learning
Used Elastic Net with feature selection combined with Modified Naive Bayes to do the prediction.
Mini-Minecraft Game Engine in Computer Graphics
• Constructed terrain using Perlin Noise and texture animation.
• Allowed interaction with terrain and accelerated rendering time by optimizing shader program.
Quaternion-Based Unscented Kalman Filter for Robust Motion Tracking
• Estimated the underlying 3D orientation by learning the appropriate model parameters from ground truth data given by a Vicon motion capture system.
Simultaneous Localization and Mapping with Particle Filter(SLAM)
• Integrated the IMU orientation and odometry information from a walking humanoid robot with a 2D laser
range scanner (LIDAR) in order to build a 2D occupancy grid map of the walls and obstacles in the
environment.
Playing Atari with Deep Reinforcement Learning
• Used deep learning model to learn control policies directly from high-dimensional sensory input using
convolutional neural network trained with a variant of Q-learning.