Australian AI Roadside Tech Prevents Wildlife Collisions
Australia is making significant strides in addressing the pervasive and often tragic issue of roadkill, a problem that claims the lives of millions of animals annually and poses substantial risks to human safety. Researchers from the University of Sydney, in collaboration with the Queensland University of Technology and the Queensland Department of Transport and Main Roads, have developed and successfully trialed a groundbreaking AI-driven roadside technology named LAARMA, or the Large Animal Activated Roadside Monitoring and Alert System. This innovative system represents a crucial step forward in road safety and wildlife conservation efforts.
Roadkill is a critical concern in Australia, with an estimated 10 million native animals, including mammals, reptiles, and birds, falling victim to vehicle collisions each year. Beyond the devastating impact on biodiversity, particularly for threatened species like cassowaries and koalas, these incidents also lead to human injuries, fatalities, and significant vehicle damage. Traditional mitigation methods, such as fencing and wildlife crossings, while effective, are often limited by geographical and financial constraints. This highlights the urgent need for scalable and adaptable solutions, which LAARMA aims to provide.
The LAARMA system is a low-cost, AI-powered roadside unit designed for real-time detection of large animals near roads. It employs a sophisticated suite of pole-mounted sensors, including RGB cameras, thermal imaging, and LiDAR, to monitor the environment continuously, even in challenging weather conditions. A key innovation of LAARMA is its self-training AI, which continuously learns and improves its detection accuracy over time without extensive human intervention. Dr. Kunming Li from the University of Sydney’s Australian Centre for Robotics explained that the system “teaches itself to get better” and “learns something new” every time it spots an animal.
Once an animal is detected, the system immediately triggers nearby flashing Variable Message Signs (VMS) to alert drivers. The messaging content displayed on these signs is carefully designed to maximize driver engagement, presenting both the type of animal detected and recommended actions, such as slowing down and scanning the road ahead. A field trial conducted over five months in Far North Queensland, a known hotspot for cassowary collisions, demonstrated promising results. The system detected cassowaries with a high accuracy of 97% and recorded over 287 sightings. Crucially, when the warning signs were activated, driver speeds noticeably decreased by as much as 6.3 km/h, significantly lowering the risk of collisions. By the end of the trial, the system’s detection accuracy for animals within 100 meters improved from an initial 4.2% to 78.5%.
The researchers are making the code powering this AI technology freely available worldwide on GitHub. This open-source approach means that conservationists and researchers globally can adapt and develop animal-specific models, potentially saving endangered species in other regions, such as red pandas in Nepal or giant anteaters in Brazil.
This Australian innovation comes amidst a broader global push for technological solutions to wildlife-vehicle collisions. Other initiatives in Australia include trials in New South Wales focusing on AI-assisted animal detection systems linked to “smart” roadside signs to warn drivers about kangaroos, koalas, and wombats. There are also ongoing discussions and research into “virtual fencing” technologies that use light and sound signals to deter animals, though their effectiveness is still under evaluation.
The development of LAARMA signifies a pivotal moment in the intersection of artificial intelligence, road safety, and environmental conservation. By providing real-time, adaptive warnings, this technology offers a tangible pathway to significantly reduce roadkill, protect vulnerable wildlife populations, and enhance safety for motorists across Australia and potentially around the world.