Tiga Leung Ho Yin

Robotics Enthusiast | Navigation and Control










Tiga Ho Yin Leung

Tiga Ho Yin Leung

MSc Student at Cranfield University studying Aerospace and Robotics

Cranfield University

Biography

Hi! I am an MSc student at Cranfield University studying Aerospace and Robotics.

My interests intersect with robotics and mechatronics, artificial intelligence and control system. I am currently working on Learning off-road vehicle traversability using Semantic Segmentation and Deep Reinforcement Learning.

Before going back to the University for my master studies, I have started my career by joining the Autonomous Driving Research team in SenseTime Group Limited as a Researcher Intern once I graduated from the Chinese University of Hong Kong in 2018. I was involved in researching and implementing different path planning and control algorithms, such as RRT-Star, Lattice Planner, Timed-Elastic-Band Trajectory Optimisation, PID control as well as Model Predictive Control. Besides, I was one of the main developers in the SenseRover Pro Project. SenseRover Pro is a 1:10 scale autonomous driving car that mimics the level-4 autonomous driving system, which was later commercialised as an educational product for different high schools in Hong Kong and Mainland China.

After working for a year, I switched to a more research-oriented job – Research Assistant at The Chinese University of Hong Kong. I was under the supervision of Prof. ZHOU Bolei in the Multimedia Lab (MMLab), which is one of the pioneering institutes on deep learning. My research focuses on Mobile Robot Navigation, Robot Arm Manipulation, and Deep Reinforcement Learning.

Interests

  • Robotics
  • Artificial Intelligence
  • Autonomous Driving
  • Control System
  • Mechatronics

Education

  • BENG in Mechanical and Automation Engineering, Minor in Computer Science, 2018

    The Chinese University of Hong Kong

  • MSc Autonomous Vehicle Dynamics and Control, 2022

    Cranfield University

Skills

Robot Operating System (ROS)

80%

C++

85%

Python

85%

Linux Operating System

90%

Computer Aided Design

70%

Passion

100%

Experience

 
 
 
 
 

Junior Research Assistant

The Chinese University of Hong Kong

Sep 2019 – Aug 2020 Hong Kong

Conducting research related to Indoor Navigation and Robot-Arm Grasping. Responsibilities include:

  • Conducting research in Robotic Grasp in the manner of Imitation Learn and Deep Reinforcement Learning
  • Researching a hybrid framework that combines learning-based navigation techniques with model-based planning and control method
  • Built up a mobile robot with ROS and PyRobot Library supported
  • Conducting experiments with AI-habitat
 
 
 
 
 

Researcher Intern

SenseTime Group Limited

Jun 2018 – Aug 2019 Hong Kong

Conducting research on Path Planning Algorithm of an Autonomous Driving Car. Participating the R&D process of a 1/10 scale autonomous driving car. Responsibilities include:

  • Developed a 1/10 scale autonomous driving car that mimic the level4 autonomous driving system pipeline as a commercialised educational product for different high school in Hong Kong and Mainland China.
  • Participated in the research work on RRT-Star Path Planning Algorithm
  • Researched and Implemented different SLAM Algorithm.
  • Developed a simulation environment for car-like robot using Gazebo.
 
 
 
 
 

Engineer Summer Internship

Chevalier Singapore Holdings PTE LTD

Jul 2017 – Aug 2017 Singapore

Responsibilities include:

  • Designed the mechanical components of an elevator system with CAD
  • Implement the BOM for the division

Projects

ROS Package cooperating OpenAI Gym and Baselines, MuJoCo Simulation environment, Gazebo Simulation and Facebook PyRobot/LoCoBot

This package is used for training and validating the Deep RL control policy on the LoCoBot, a low-cost robot equipped with a 5 DOF robotic arm on top of a differential-drive mobile base. Deep RL Policy can be trained and validated in both MuJoCo and Gazebo Simulator.

SenseRover Pro

SenseRover Pro is 1:10 scale autonomous driving car that miniaturing the level-4 autonomous driving system of a 1:1 autonomous driving car. It is equipped with all sort of sensors, actuators, and on-board computer which mimics the setup of a 1:1 autonomous driving car. System Pipeline and software design is much similar to a full scale solution. It is a commercialised educational product for different high school in Hong Kong and Mainland China.

Autonomous RobotCar

RobotCar is a 1/10 scale Autonomous vehicle prototype. It carries a suite of sensors, perception, planning, control modules that are similar to a full scale solutions. Compare to others prevalent commercial research robots like Jackal UGV or TurtleBot, RobotCar address the issues of realistic vehicle dynamics; Ackermann steering and the ability to travel at a high speed are the main difference among other mobile robot platforms.

Autonomous Lane Tracking RaceCar

This project is my first project during my internship in SenseTime Corporation. It involved developing a 1:10 scale autonomous driving car capable of tracking double lane racetrack in real time and simple object detection. Solutions powered by OpenCV and Tensorflow had been adopted.

Recent Posts

Recent Publications

Cross-view Semantic Segmentation for Sensing Surroundings

Sensing surroundings plays a crucial role in human spatial perception, as it extracts the spatial configuration of objects as well as …
Cross-view Semantic Segmentation for Sensing Surroundings