Science & Technology, Australia (Commonwealth Union) – Australia is a continent prone to bushfires due to a combination of factors. It’s hot and dry climate, vast expanses of dense, flammable vegetation, and strong winds create the perfect conditions for wildfires to ignite and spread. In fact, many Australian plants have evolved to rely on fire for their reproduction, which further complicates the situation.
Technologies like fire weather forecasting, satellite imagery, and alert systems are used to provide early warnings to residents are currently in use and the addition of AI may boost the battle against bushfires.
Researchers have revealed that snapping photos during your next bushwalk has the potential to play a vital role in predicting and preventing devastating bushfires. This innovative mobile phone application, named NOBURN, incorporates cutting-edge AI technology developed by experts from the University of Adelaide.
According to Professor Javen Shi, the Director of Advanced Reasoning and Learning at the University’s Australian Institute for Machine Learning (AIML), indicated that the app’s AI-powered computer vision can evaluate images submitted by users, depicting fire-prone areas, in order to gauge the potential fuel loads for bushfires. The data submitted is then utilized to estimate the severity of a potential bushfire and predict its likely extent.
Professor Shi elaborated on how the AI models within the app emulate the visual acumen of human experts who traditionally survey bushlands for potential bushfire hazards.
“Once we train the model, we don’t really need experts to go to the forest,” added Professor Shi.
“We can easily have thousands of citizens – such as bushwalkers and families on camping trips – who can take photos that our AI can use to make predictions.”
Gathering high-quality data from NOBURN users may require a timeframe of up to 12 months, with an additional year necessary for the app’s AI model to undergo training using the images submitted from rural areas.
Nonetheless, Professor Shi emphasized that the primary purpose of the app is to promote awareness of the potential of AI investment in preventing catastrophic losses caused by bushfires. He indicated that while the app itself may appear modest in scope, their aim is to spark enthusiasm and optimism as each year, they suffer tremendous financial losses due to wildfires.
He further articulated the app’s broader vision, indicating that their aspiration is to expand its functionality to encompass situational awareness, serving as a command center for bushfire commanders. This would enable real-time monitoring of resource deployment, tracking the spread of bushfires, and the development of interactive AI tools to assist them in their decision-making processes.
“We even foresee an AI that can have a conversation with the commander and the firefighters – like how Iron Man has AI talking to Tony Stark – to give real-time advice and make them more aware of their surroundings in low-visibility conditions.”
The concept for the NOBURN app originated in the wake of the devastating Black Summer bushfire season in 2019-2020. During this season, more than 46 million acres were consumed by fires, resulting in the destruction of over 9,000 properties, the loss of 34 lives, and staggering financial losses of up to $7 billion for farmers.
To bring this innovative idea to life, Professor Shi collaborated with forestry experts from the University of the Sunshine Coast, including Dr. Sam Van Holsbeeck and Professor Mark Brown. Additionally, Professor Paul Salmon, the Director of the Centre for Human Factors and Sociotechnical Systems, played a key role in the development of the NOBURN app.
Dr. Van Holsbeeck emphasized the immense potential of AI in the realm of fuel-hazard assessment and its further applications in bushfire prediction and simulation modeling, aligning with the research conducted by AIML. He emphasized that the NOBURN project transcends typical bushfire research, combining various fields of expertise to make bushfire research accessible to the general public.
