The Queensland Department of Transport and Main Roads has partnered with researchers from the Central Queensland University (CQU) to develop a new automated system to track and analyse road safety conditions using cars fitted with laser scanners.
The project will adapt "deep learning" image pattern recognition software to detect and classify roadside objects and assess whether their proximity to the road is hazardous.
It comes as governments across Australia grapple with the imminent arrival of autonomous vehicles which use hazard detection systems to avoid collisions and mishaps.
The Centre for Intelligent Systems at CQU will use vehicle-mounted cameras and laser scanners to survey conditions and collect road data for the project.
Jointly funded by an Australian Research Council Linkage grant and the Queensland government, the project is part of efforts to slash the high number of Australian road fatalities which, on average, claim around 100 lives a month.
Project leader and director of CQU’s Centre for Intelligent Systems (CIS) professor Brijesh Verma told iTnews developments in deep learning and image processing will improve on the current manual systems for judging road safety, which are inefficient and prone to errors.
“The final outcome will be new identification techniques and software which can be incorporated with [current] road data collection systems,” he added.
The classification of roads and hazards will then be identified according to attributes listed in the Australian Road Assessment Program (AusRAP) run by the Australian Automobile Association.
Verma said the new project, which will run for three years, will collect data on Queensland’s roads every year and “has the potential to provide a range of value-added products for safer roads”, namely to reduce fatal collisions and improve asset management.
“Automatic assessment of road safety and conditions is essential for improving road infrastructure and reducing fatalities on the roads,” he said.
The detection and cataloguing of hazardous road conditions and nearby dangers is part of a wider data-driven approach to fixing roads that uses hard evidence to inform decisions and priorities.
