Astri Logo White
Search icon
Astri Logo White
search icon

Detection of Human Interaction for Visually Impaired People (ARD/317)

Project Title:
Detection of Human Interaction for Visually Impaired People (ARD/317)
Project Reference:
ARD/317
Project Type:
Seed
Project Period:
03 / 04 / 2024 - 02 / 04 / 2025
Funds Approved (HK$’000):
2,794.500
Project Coordinator:
Mr Wai Cheong KU
Deputy Project Coordinator:
/
Deliverable:
Research Group:
/
Sponsor:

Hong Kong Blind Union

Description:

According to the statistics, Hong Kong currently has around 120,000 visually impaired people. People with visual impairment are in fact no different from normal people besides the impairment. However, there are public facilities that require visual interpretation of the social roles of specific personnel to utilize. These give the visually impaired people a lot of difficulty. The objective of this project is to develop the technology for enhancing the social activities of the visually impaired people, especially the social contact and interaction with the normal people. In this project, ASTRI will design five algorithms. The first three algorithms are to visually interpret human interaction for the visually impaired people, one for finding passerby people for enquiry, one for recognizing individual social roles and one for recognizing the behavior of a group of people. The fourth algorithm and fifth algorithm are human-environment interaction. The fourth algorithm is to recognize public facilities that are inaccessible without prior knowledge of their existence. The fifth algorithm is to provide safe navigation by avoiding obstacles in the immediate vicinity. To realize the algorithms, we will need some supporting hardware, i.e., lidar, cameras, IMUs and an autonomous robot. These components are compactly available given a quadruped robot. We will use a quadruped robot as the hardware platform to utilize verbal commands, voice feedback and navigation feedback to address the unique challenges for being visually impaired. The navigation feedback is delivered to the user via a rigid handle connecting the autonomous robot. Lastly, a trial will also be conducted to verify our algorithms.

Co-Applicant:
/
Keywords:
/