Using AI on the fireground

Source: Victoria Country Fire Authority

Artificial Intelligence (AI) is the buzzword of the moment, promising to revolutionise everything from writing emails to driving cars.

But for CFA the adoption of new technology isn’t about following trends. When lives and property are on the line, our systems cannot simply be ‘smart’; they must be robust, reliable and proven in the harshest conditions Victorian summers can throw at us.

While CFA takes a cautious approach to integrating AI, recognising that technology must never replace the skill and judgement of our people, we also recognise its potential to solve long-standing operational challenges.

The case for change

The push for smarter technology is driven by real-world needs. For example, we know that removing crew members from the back deck of tankers significantly reduces the risk of injuries from burnovers, falls and heat exhaustion. Remote-controlled monitors have been the first step in this evolution, allowing crews to fight fires from the safety of the cabin.

The effectiveness of bullbar-mounted monitors was proven during the Boxing Day 2024 fires in the Grampians. Crews operating heavy tankers with monitors played a pivotal role in halting a fire run that threatened the township of Moyston. Feedback from the incident highlighted that these tankers and their monitor capability were instrumental in saving homesteads and historic structures.

However, current remote monitors can be difficult to aim accurately while a vehicle is bouncing over rough terrain during ‘pump-and-roll’ operations – a dynamic tactic where the tanker simultaneously drives and applies suppression. This is where AI enters the picture.

From desktop to the real world

For the past few years, CFA has partnered with Deakin University’s Institute for Intelligent Systems Research and Innovation (IISRI) on a feasibility study to assess whether low-cost cameras and algorithms could identify fire edges and heat signatures to assist with targeting monitors.

Desktop testing confirmed the concept is sound. We know the cameras can ‘see’ the fire, but a lab environment is different from a fireground. To trust an AI system with targeting water or identifying hazards, it needs to learn from the chaotic reality of smoke, dust, vibration and glare.

This need for high quality data drove recent field testing initiatives, including data collection at hazard reduction in Swan Hill. More than 700 gigabytes of data was collected in just 90 minutes of burn from three different options of thermal infrared cameras that can be mounted to tankers. This data can be used to train the AI to distinguish between a burning fence post, a person and a spot fire, ensuring the system works when it matters most.

Scaling up 

We are now entering an exciting new phase. The project has successfully secured funding from the Australian and New Zealand Council for fire and emergency services (AFAC) via the ‘Bushfire Emerging Technologies Hub,’ in collaboration with Google.

This next stage will move beyond theory to field tests and scale AI object detection and targeting algorithms. The project aims to refine the technology to support safer pump-and-roll firefighting.

The new funding will allow CFA to:

  • Enhance safety: Test performance of hazard detection capabilities to identify pedestrians, fences, trees and rocks near vehicles, preventing collisions during low-visibility or high-stress conditions.
  • Improve efficiency: Train the AI to classify fire edges in real-time, helping to automate or assist nozzle targeting so crews can focus on fire behaviour rather than fighting the joystick.
  • Work offline: Optimise software to run on ‘edge computing’ systems local to the truck, so that no internet connection is required for the AI to work.
  • Feed intelligence: Send simplified ‘fire/no-fire’ data back to support mapping and improve situational awareness without clogging radio networks.

By combining the skill of our firefighters with the precision of modern technology, CFA seeks to chart a course for safer tankers in the future where our crews can do their job more effectively and, most importantly, come home safely.

Submitted by Nick McCarthy