Projects
LiDAR Point Cloud S2D (GAN)
2019.09Built a GAN model based on Pix2Pix to transform point cloud projection images into camera depth images. Enables transformation between point cloud and image coordinates.
DARPA SubT Challenge
2019.08Team NCTU (7th Place). Developed autonomous mapping and navigation for underground environments using multi-agent systems.
Marine Robot Tracking
2019.03Tracking and following marine robots using SSD (Single Shot MultiBox Detector), Kalman Filter, and PID control with visual and point-cloud data.
Point Cloud Classification
2019.01Deep learning classification of 3D point clouds using 2D projection images. Deployed in RobotX 2018 competition.
Maritime RobotX Challenge
2018.123D Perception Leader (5th Place). In charge of 3D LiDAR and depth camera perception, including detection and classification.
NCTU Self-Driving Robot
2018.06Built a self-navigating robot for campus environments using 16-bin LiDAR, IMU, GPS, and wheel odometry.
3D Object Modeling
2018.01Built a 3D point cloud modeling system using PCL, ICP, iSAM, and apriltags with a single depth RGB camera.
Multi-robot Patrolling
2017.12ROS-based multi-robot patrolling system for surveillance and logistics applications.
