Should AI developers focus on helping firefighters?

It seems that a new breed of artificial intelligence (AI) software can learn about and predict how flames will react.
29 October 2018

Can AI really help firefighters? Source: Shutterstock

Fires are unpredictable, and there is practically no data about how they spread. However, its possible to both, create and capture data about fires and fit that data into algorithms to help firefighters do a better job, a safer job.

Fortunately, it’s something that developers, engineers, and Fire departments are already exploring, with help from serious government agencies.

Last month, for example, the Cosumnes Fire Department ran a test. It’s crew reported to a series of fully-involved structure fires in Elk Grove, California — with limited ventilation.

Teams were assigned to extinguish a fire that was set up with ordinary and modern combustibles and outfitted with a complex array of thermocouple sensors, advanced thermal and visual imaging devices, and heat-resistant 360-degree cameras.

The objective of this exercise was to gather data to train AUDREY, also known as Assistant for Understanding Data through Reasoning, Extraction and Synthesis, a software program created by National Aeronautics and Space Administration’s (NASA) Jet Propulsion Laboratory (JPL) with monetary support from the Department of Homeland Security’s (DHS) Science and Technology Directorate (S&T).

The software is designed to perform data fusion and provides tailored situational awareness to first responders.

The series of “test burns” that the Cosumnes Fire Department were subjected to were aimed at providing mountains of data, sensor readings, and imagery to begin training AUDREY on fire growth, flow paths, and flashover.

“Leveraging cutting-edge technology from the newest field of artificial intelligence (AI), Artificial General Intelligence (AGI), AUDREY is being taught fire behavior and the risks firefighters face to assist firefighters, incident managers, and dispatchers to keep personnel safe,” experts explained to FireEngineering.

Each scenario that the Cosumnes Fire Department faced while putting out the test fire was aimed at teaching AUDREY how fast a fire develops, how the contents of the fire affect the heat and the fire’s growth, and how much time responders have to do their job before conditions are no longer tenable.

“AUDREY takes video data and sensor data and fuses it together to determine what the fire is going to do next. The data we are capturing will be fed into these models, machine learning will take place, and we will get some understanding of what’s happening and [how to] develop prediction technologies [so] we can look into the future,” explained FLAIM Systems’ expert Dr. James Mullins to the media.

Although AI, or AGI as we’re now learning, isn’t here to help firefighters or first responders as yet, the mere germination of the idea suggests that we’re getting close to building something meaningful — something that will save the lives of real people someday.

To enterprises that make the right investments in this sphere, the rewards aren’t limited to financial gains but also recognition in society.