Projects

Group Projects

  • Coustomized RL Enviornment for daVinci Sugical Robot
    1. Coustomized a gym-like RL enviornment for daVinci Surgical Robot
    2. Implemented DDPG and TD3 for reaching task
    [report] [poster]

  • Inner-Class and Inter-Class Style Transfer using CycleGAN
    1. Explored the domain gaps between inner-class domains and inter-class domains
    2. Implemented CycleGAN 1 and trained it on different data
    [report] [presentation]

  • Autonomous R/C Vehicle
    1. Built a remote control vehicle that can autonomously run on an outdoor scaled track
    2. Improved the vehicle by adding traffic signs recognition functionality and speeding up the video processing efficiency three times using the multi-threaded approach
    [report] [presentation]

  • Deep Learning for Plankton Image Retrieval
    1. Developed a content-based image retrieval system for plankton images using a deep convolutional neural network which assisted biological oceanographers in researching and labeling the plankton images
    2. Researched on different machine learning and deep learning methods, which accelerated the searching process and improved the precision of the baseline retrieval system by about 30%
    [presentation] [video]

  • Drone Integration for RF Scanner Payload
    1. Integrated an RF scanning payload with a drone (DJI Matrice 100) to automate the processes of detecting wireless signal's strength in open area
    2. Developed a mobile app to record the signal strength data and generate the heatmap which can visualize the data better
    [report] [presentation]

  • Electromagnetic Cannon
    1. Built an electromagnetic cannon with $100 budget
    2. Implemented circuit design, 3D printing, Arduino coding and make distinct functions to work together to complete a bigger task
    [report] [presentation]

Computer Vision

Intelligent System and Robotics

  • Color Segmentation, Logistic Regression [report]

  • SLAM with Partical Filter [report]

  • SLAM with Extended Kalman Filter [report]

  • Shortest Path with Dynamic Programming [report]

  • Optimal Strategy with Dynamic Programming [report]

  • A*, LRTA*, RRT* [report]

  • SARSA and Q-learning [report]

  • Value Iteration and Policy Iteration [report]

  • Model-based Control [report]

  • VI, PI and Q-learning [report]

  • Policy Gradient [report]

  • Deep Deterministic Policy Gradients [report]

Others