Large Australian animals and cars: Safety for all
The issue of animal-vehicle collisions (AVC) looms large in Australia, resulting in injuries and fatalities to both. As roads increasingly encroach on habitat areas, there are regular instances of animals moving across road corridors. Fencing corridors incurs negative impacts on animal population distributions and are expensive to maintain.
The development and testing of a system to help lessen the danger was the focus of the recently-completed Large Animal Activated Roadside Monitoring and Alert System project. Final reports from the work are available for download below.
Partners with iMOVE on this research were the Department of Transport and Main Roads (Queensland), the University of Sydney’s Australian Centre for Robotics, and the Queensland University of Technology’s CARRS-Q.
Project objectives
- Accurate and reliable detection of large animals at distances of up to 200 metres, in various weather conditions (daytime, nighttime, rain, and dry weather). The system would use a machine learning pipeline to improve detection accuracy over time.
- The system would be designed to train itself, using data collected in the field and supplemented with synthetic data. This self-supervised learning approach would reduce the need for extensive human intervention.
- Open-source software would be developed to allow the system to be easily adapted and deployed in new locations.
- Evaluation of changes in driver behaviour in response to real-time messaging alerting motorists of an animal on or near the road as displayed on the roadside variable message sign (VMS).
The ultimate goal was to provide a scalable, adaptable system capable of enhancing road safety and promoting wildlife conservation, particularly in areas with a high prevalence of AVCs.
Methodology
Messaging concepts
Before commencement of any simulated or real-world driving trials, a series of roadside Variable Message System (VMS) was developed and then concept-tested firstly via qualitative focus groups, and then on a larger scale using an online survey with over 550 participants.
Results of testing the messaging were that participants felt it was more effective when:
- the type of animal be identified, the specificity of the message increase the likelihood of drivers to take the warning seriously
- the ‘slowing down’ message was the first message displayed,
- that it was plainly understood by drivers that this was a real-time warning, and that if signs were only activated when animals were detected in the area. To do otherwise, i.e. leaving the signs on permanently, would likely result in driver complacency.
Driving simulator study
Next was a driving simulator study, involving 51 participants on a 40-minute simulated drive, in which two of the VMS messages are encountered. Participants were exposed three times to two similar scenarios where cassowaries were either crossing the road, walking along the roadside or not appearing.
Two key driving reactions were assessed:
- The approach zone, defined as the 5 seconds before reaching the VMS.
- The event zone, defined as the area immediately surrounding the detected cassowary
The simulator study found that drivers responded positively to the VMS messaging by reducing their speed in the approach zone. However, speed reductions in the event zone were less significant, possibly because participants were aware that the simulation did not present a real risk of collision with a cassowary. Nevertheless, the study provided valuable insights into how drivers interpret and respond to VMS messages and helped to provide further validation of the messaging content for use in the subsequent field trial.
Field trial in Far North Queensland
The final part of the project involved a five-month field trial in FNQ, at a site with known cassowary activity. Three months of the trial were dedicated to data collection for the model training, flowed by two months of on-road trialling of the LAARMA system. During the five months there were 287 verified cassowary sightings recorded.
The field trial assessed not only driver reaction, but also the effectiveness of the LAARMA system in detecting the presence of animals.
In the case of driver reaction, the system had the following results:
- vehicle speeds in event zones at two test sites decreased by an average of 6.30 km/h and 5.06 km/h respectively, corresponding to an almost 10% reduction to the posted speed limit
- in approach zones there were reductions of 4.26 km/h and 3.44 km/h at the two zones.
In regard to the systems success in detecting animals:
The results showed that the system achieved a precision of 0.77 and a recall of 0.97 during the on-road trial. This means 77% of the events the system triggered involved cassowaries, and the system accurately triggered for 97% of the events where cassowaries were present.
Future directions
The results from the FNQ trial were promising and given the ever-present danger resulting from incidents between vehicles and large, uniquely Australian animals, there is definitely a need to take the work further.
Firstly, although this project had a particular focus on the detection of cassowaries, the system was designed and developed with the capability to extend its application to large animals more broadly. This could involve training the machine learning models on more diverse datasets and conducting field trials in different environments.
It was also recognised that there could be some improvement in the size, array, and placement of sensors, in order to improve accuracy and increase detection range.
And although the VMS used was successful in the cassowary trial, the content displayed would need further research, as “there may be differences in how drivers respond to warnings about different species.”
Finally, “as the system can be easily deployed on different sites, the long-term effect of the system must be evaluated, in order to understand what strategies can be employed regarding the deployment of the VMS at multiple sites.”
Conclusion
“The project’s success in reducing vehicle speeds and validating the system’s effectiveness in real-world conditions provides a strong foundation for future research and development aimed at refining the system and expanding its deployment to new regions and potentially new animal species.
Looking ahead, the LAARMA system has the potential to play an important role in reducing AVCs and promoting wildlife conservation, not just in Australia but also in other regions facing similar challenges. By addressing the recommendations outlined in this project, the LAARMA system could be further refined and scaled to contribute to improved road safety and wildlife conservation efforts both within Australia and internationally.”
Expected project impacts
The outcomes of this research are very promising. To have a low-cost system that can be deployed at specific locations where there is a history of Vehicle-animal collisions and be able to improve the awareness and advanced warning to drivers would provide a road safety improvement.
Ross Hodgman, Department of Transport and Main Roads Regional Director (North Queensland)
Download the final reports
There are two reports available for download:
- Final Report: The Development and Performance Testing of a LAARMA—Large Animal Activated Roadside Monitoring and Alert System
- Project Summary: The Development and Performance Testing of a LAARMA—Large Animal Activated Roadside Monitoring and Alert System
Also in the reports are QR codes, making the project source code freely available via GitHub.
Webinar video
On 29 July 2025 we ran a webinar investigating the outcomes from this project. The event was recorded, and is available to watch at: Safer roads for humans & animals: Proving the LAARMA System
UPDATE: September 2025
An academic paper based on this project’s work has been published in IEEE Robotics and Automation Letters (Volume: 10, Issue: 10, October 2025), one of the top journals in robotics area.
The paper is entitled Endangered Alert: A Field-Validated Self-Training Scheme for Detecting and Protecting Threatened Wildlife on Roads and Roadsides.
Discover more from iMOVE Australia Cooperative Research Centre | Transport R&D
Subscribe to get the latest posts sent to your email.