PROJECTS

University of California, Santa Cruz

Master Thesis – LIDAR A* (9/2019 – 6/2020)

  • Advisors: Roberto Manduchi
  • Researched for a new efficient method for real-time mobile robot motion planning in an unconstructed environment. LIDAR A* features an online visibility-based decomposition and search process that incrementally analyzes regional maps with locally simulated 2D LIDAR scans to search for the shortest gap sequence toward the goal. The gap sequence is used to guide the simulation of a local planner to generate a smooth and kinematic-friendly path. The significant node reduction minimizes computation time, allowing the algorithm to produce a near-optimal path on a large and highly complex map in real-time [C++]
  • Link to Thesis (ResearchGate)

  • Each frame is computed from scratch without any pre-processing or memorization

The “Slugnificent” Seven (5/2019 – 6/2019)

  • Built a mobile robot that can autonomously travel to an initial firing zone by following tape, avoid obstacles with bumpers, and use beacon to locate and shoot ping-pong balls at the opponent robot [Embedded C]
  • Developed a tape sensing module that determines tape orientation and offset relative to the robot, which helps with creating smoother robot motion
  • In charge of system design and entire software development of the robot

Cat Mario AI (1/2019 – 3/2019)

  • Developed a Cat Mario AI based on NeuroEvolution of Augmenting Topologies (NEAT) method that uses genetic approach to learn to play the game of Cat Mario. Cat Mario is a parody game of the Super Mario Bros known for its level of difficulty that causes extreme frustration [Python]
  • Utilized classic computer vision to capture real-time video feed, classify objects (cat agent, enemies, and neutral platforms), and convert game frame into low dimensional training data
  • Mapped the game level by tracking the movements of the side-scrolling game window and various in-game static objects in real-time
  • Developed a training environment that allows the genomes (neural network agents) to play the game, identify game events, and show trajectories and training statistics
  • Link to Poster

Autonomous Ping-Pong Robot Simulator (4/2018 – 6/2018)

  • Individually developed a 3D simulator to simulate gameplay of an autonomous mobile Ping-Pong Robot. System includes a paddle mounted on a 6-axis robot arm on an omnidirectional mobile robot platform [LabVIEW]
  • Built a full 3D environment with LabVIEW 3D graph to generate and transform models of robotic arm, omni robot platform, paddle, and ping-pong table
  • Developed physics models for ping-pong ball to interact with the table, the net, and moving paddle
  • Determined forward and inverse kinematics for 6-axis robotics arm and mobile platform to create a strike that lands on the opponent’s side of the table

  • Due to time constraints in development, the robot is teleported to its target pose to strike the ball
  • You may turn on CC for action descriptions

California State University, Northridge

Brain-Computer Interface (BCI) Wheelchair Project (9/2011 – 4/2014)

  • Developed an intelligent wheelchair that analyzes the user’s facial gestures and brain signals from an electroencephalogram (EEG) headset to generate motion commands. It utilizes GPS, LIDAR, and computer vision to autonomously navigate around the campus [LabVIEW]
  • Developed computer vision software that identifies areas of pavement, grass, dirt, and shadow under outdoor light conditions
  • Coded the intelligent self-driving system that controls the wheelchair to travel to a set of GPS waypoints on campus

Intelligent Ground Vehicle Project – Red RAVEN (1/2010 – 1/2013)

  • Developed autonomous mobile robots and competed with ~50 teams each year at AUVSI’s international Intelligent Ground Vehicle Competition (IGVC) [LabVIEW]
  • 2010 5th place, 2011 and 2012 repeat 1st place Grand Award at IGVC
  • Provided demo at NASA Jet Propulsion Laboratory (JPL) and at AUVSI Unmanned Systems North America 2012 exhibition
  • Recipient of Distinguished Engineering Project Achievement Award 2013 given by the San Fernando Valley’s The Engineers’ Council
  • Coded LIDAR and classic computer vision perception that detects construction barrels, lane boundaries, flags, and creates real-time 3D visualization on the GUI. Rewrote local motion planning algorithms for obstacle avoidance
  • Design Report: 2011, 2012