Auslan Assist: Automated system for public transport

Auslan Assist aims to develop and trial Australia’s first real-time Australian Sign Language (Auslan) translation system for public transport announcements, using a digital avatar. The system will translate text into Auslan and vice versa, the preferred language of the Australian Deaf Community, ensuring equitable access to transport information.

The project will also make significant progress toward enabling automated translation of Auslan into English and vice versa.

.

Participants

• Transport for NSW
• Queensland University of Technology

Project background

Automated audio announcement and visual systems are widely used across public transport systems in Australia. These systems may not be fully accessible to customers who cannot hear audio announcements or who cannot understand the English text displayed on digital screens.

Prior research by QUT has found that lack of access to timely, crucial information while travelling is a barrier that prevents many deaf Australians from using public transport.

This project aims to develop and trial Australia’s first real-time Auslan translation system for public transport announcements, using a digital avatar. Working with native signers from the Australian Deaf Community, the project will use motion capture to produce a library of sign animations tailored for transport use and able to be assembled into announcements in real time.

The project will also make significant progress toward enabling automated translation of Auslan into English and/or English into Auslan, through creation of an AI model capable of recognising Auslan signs. The system will be co-designed and evaluated with members of the Australian Deaf Community, Deaf Australia, and Deafness Forum Australia, to ensure strong user acceptance and uptake.

The project builds on substantial prior collaboration between QUT and Transport for NSW, including the anticipated successful completion of a trial of a limited system on the Sydney Trains network.

Project objectives

1. Develop an automated system to generate real-time Auslan public transport announcements via a digital avatar to integrate into Sydney Trains systems.
2. Integrate the Auslan Assist system with Sydney Trains’ existing announcement infrastructure for trials.
3. Evaluate the effectiveness and community acceptance of the system among the Deaf community and sector.
4. Explore and evaluate how machine learning gesture recognition can be adapted for bi-directional translation, with the goal of progressing two-way translation between Auslan and English to proof of concept.

Please note …

This page will be a living record of this project. As it matures, hits milestones, etc., we’ll continue to add information, links, images, interviews and more. Watch this space!