Australian Driverless Vehicle Machine Learning Dataset Project
Automated vehicle (AV) technologies can provide benefits to the existing transport ecosystem such as increased road safety, passenger mobility and connectivity. As part of Transport for NSW’s research and development to prepare for the deployment of AV technologies, this Dataset is being collected to foster the development of the local CAV industry and address the gap in dataset coverage in Australasia. This initiative aims to:
- Lower the barriers to entry for CAV software developers, industry and academia
- Form the basis of AV test beds tailored to the diversity of Australian road conditions
- Create the first Australian large-scale open CAV benchmarking dataset
- Address priority areas identified through interviews and surveys with key stakeholders, industry and community groups
- Align with the priorities and goals of the Connected and Automated Vehicle (CAV) Readiness Strategy and the Future Transport Strategy 2056
Transport for NSW are producing this AV dataset in collaboration with the National Transport Research Organisation (NTRO, formerly the Australian Road Research Board (ARRB)) and the Australian Centre for Field Robotics (ACFR).
Data collection and annotation was undertaken in September 2023. The Dataset is currently undergoing post-processing and annotation.
The data collection vehicles were driven to and around major cities, regional towns, and rural areas including:
- Central Coast Region
- Blue Mountains Region
- Batemans Bay
The Dataset will contain around 1000 20 second long scenes, annotated with metadata and downloadable with supporting documentation.
Scenes were captured from across metropolitan, regional and rural NSW with a wide range of weather and traffic conditions represented, using two different vehicles. A detailed route will be included in the supporting documentation.
Last updated: November 2023