Team:

Hyung Woon is a 4th year undergraduate student at the NUS School of Computing, majoring in Computer Science and Psychology. He has experience with Computer Vision-based Machine Learning and AI, and his role was to design and implement the algorithm for the Robot Dog to detect and make decisions on navigation obstacles such as staircases.

 

Zenith is a 3rd year undergraduate student at the NUS School of Computing, majoring in Computer Science. His role in the project was to develop a stair climbing model.

​

Project overview:

Building on a previously developed navigation algorithm, the team designed, developed and trained a sophisticated AI pipeline for detecting and climbing staircases.

​

Why the team applied for the CCSGP fellowship:

The team believes that their goal, that is using state-of-the-art technology to improve the life quality of visually impaired people in the community, aligns well with CCSGP’s mission of encouraging the IT community in NUS to give back to society.

​

​

​

​

​

​

​

​

​

​

​

​

​

What the team learned from the CCSGP experience:

Participating in this project gave the team an opportunity for a deep-dive into how those who are visually impaired navigate around everyday obstacles that we take for granted. Designing an algorithm for detecting and climbing staircases required the team to balance between the robot's capabilities and the needs of the visually impaired community, and allowed them to approach development of Machine Learning algorithms from a new, human-focused approach. Through feedbacks from trials with those who are visually impaired, the team hopes that the robot dog will be able to improve its navigation capabilities to perfectly meet the needs of the users.

 

About the project:

This project builds on a prior project to develop a navigation technology for a robotic guide dog that helps people with visual impairment lead a more independent life, and focuses specifically on some of the more common obstacles that require special attention, such as staircases and roadside curbs. The robotic dog is able to navigate in any environment and can communicates with the user through the leash or by voice. It is a multi-function companion that can be owned by individuals, public facilities, community centers, etc.

 

The team spent weeks designing and implementing the algorithm, guided by the feedbacks from trials with individual visually impaired users, in order to better suit their needs. The collected feedbacks indicated that many of them found the robot dog immensely helpful due to it being easy-to-use, reliable, and high-tech. They were excited for this product to come to market so that they can experience its full functionality in daily life.