Home Projects

Projects and Experiences


Project Host and Smart Contract Developer for NFT “Katana N’ Samurai” (2021)

  • This project has been prepared for half a year and is divided into two generations. The first generation will be sold in September 2021, a total of 2,000 samurais 0.05 ETH each have been sold and the second generation will go on sale in January 2022.
  • The works are mainly in Ukiyo-e style. It is a traditional art famous in Japan back in 1700s, and we aim to bring this kind of historic art into the whole NFT market and spread Japanese culture.
  • I am one of the three core members of this project, mainly responsible for technical development and marketing discussions such as roadmap. The development projects are mainly ERC-721 smart contracts and front-end and back-end concatenation tasks.
  • Smart Contract: NFT art to Ethereum chain, design of NFT samurai changing clothes mechanism, whitelist purchase mechanism.
  • Front-end and back-end concatenation: use Fauna database to store metadata, front-end web buttons to interact with smart contract.

    • Github Twitter Opensea


    Business Card on the Ethereum Chain through the ERC-1155 Smart Contract (2021/12)

  • I programmed to deploy business card information of the sandbox designer to the Ethereum chain through ERC-1155 smart contract.
  • Achievement information is permanently saved.
  • Allow the owner of the business card to arbitrarily send it to a specified person or the user can claim the business card to the owner.

    • Github


    SubT Challenge Urban Circuit in Seattle (2020/02)

  • In the Urban Circuit, 10 teams navigated two courses winding through an unfinished nuclear power plant in Elma.
  • Breakthrough technologies in autonomy, perception, networking, and mobility for underground environments.
  • Team NCTU represent Taiwan to go for this competition.
  • We got 9th place in this competition.


    • SubT Challenge Tunnel Circuit in Pittsburgh (2019/08)

  • The DARPA SubT Challenge is a competition which seeks novel approaches to autonomous rapidly map, navigate, and search artifacts in an unknown underground environments during time-sensitive combat operations.
  • Team NCTU represent Taiwan to go for this competition.
  • This is also not only ours but also Taiwan's first team who participate in DARPA competition.
  • We got 7th place in this competition.


    • Pyrobot-Pick and Place Mission (2020/06)

  • Identify the object and pick it up, then move to the target location place the item.
  • Task 1: Object Detection-FCN
  • Task 2: Pose Estimation and Pick
  • Task 3: Move to destination-A*
  • Task 4: Place in the box

    • Github

    PDF


    Track and Trail Using A* Algorithm (2020/04)

  • This experiment uses LoCoBot from Facebook Research to follow the track and trail in a real environment.
  • Image recognition: MobileNet SSD for ientifying whether it is an item to be tracked.
  • Path planning: Use the A* search algorithm as the basis to control the robot's movement.

    • Github

    PDF


    Thermal and RGB Calibration (2020/01)

  • When using RGB Camera and Thermal Camera at the same time, you need to align the two sensors together.
  • Stack the thermal image on the RGB image.

    • Github


    Artifact Search Using GAN FCN-Pix2Pix-ERFNet (2020/01)

  • Sensor: RGB, Depth, Thermal
  • Use Pix2Pix-GAN concept to improve noise points of FCN model.
  • Add thermal information to increase predicted IOU.

    • Report

    PPT

    Github


    Obstracle Avoidance Using mmWave Sensor (2019/12)

  • The goal is to replace Lidar with mmwave sensor.
  • Use one or more millimeter wave radars to collect depth information.
  • Multiple millimeter wave radars collect information in a 360-degree panoramic view.
  • Implement obstacle avoidance with deep reinforcement learning method.
  • Use real and virtual methods for data collection and training.

    • PDF

    Github


    Imitation Learning Using AlexNet (2019/12)

  • Use the concept of imitation learning to complete lane following.
  • Divide the forward angle into 15 equal parts and use image to predict and classify.
  • Use AlexNet classifier.

    • Github


    Artifact Search Using SSD Algorithm (2019/08)

  • Used in SubT-Tunnel competition environment.
  • Find the target and accurately report the location.
  • Compare the advantages and disadvantages of SSD and MobileNet-SSD algorithms.

    • PDF

    Github(SSD)

    Github(MobileNet-SSD)


    Nuclear Disaster Strain Robot (2019/06)

  • This plan is aimed at the search and rescue nuclear disaster sites and areas inaccessible to humans.
  • Use the MobileNet SSD algorithm to identify objects in the field.
  • Use UWB positioning technology to get the location and mark the identified objects on the map.

    • PDF

    Github


    Duckiebot Vehicel Following (2018/12)

  • Combine Duckiebot with Deep learning model.
  • Use the CNN architecture to train image, V, and ω.
  • Able to follow the vehicle in front to change lanes and maintain a fixed distance.

    • PDF

    Github


    Duckietown @ NCTU, Autonomy Education and Research Platform (2018/08)



  • Duckietown Summer School is a course for about one hundred students from Koera, Indonesia, and Taiwan.
  • I performed as a teaching assistant during the course for a summer vacation.
  • From this experience, I learn a lot of skills such as ROS (Robot Operating System), Python, OpenCV.

    • Duckietown Summer School Website