back to article AI could save future firefighters from deadly flashover explosions

AI could help save firefighters' lives by predicting fire flashovers before they occur, according to new research published this week.  Flashovers occur when combustible material in a room suddenly starts igniting all at once, leading to a huge surge of heat and flammable gases that can break walls and burst windows. Around …

  1. claimed


    Predictions based on simulated data, requiring the model to know the building layout.

    This seems to be decades away from ever being possible to integrate.

    Perhaps instead they should look to the hardware, if firefighters had a wearable device which mapped rooms as they entered them (would lidar even work in a hot, bright room?), and gathered atmospheric info - then the realtime work was done and the device itself then altered the firefighters - it might be more achievable to save some lives.

    How many fully mapped smart buildings contribute to the 13% of incidents that result in death or injury?

    I don't mean to be too critical but this sounds like a non starter

    1. Bartholomew Bronze badge

      Re: Approach

      Would lidar even work in a hot, bright room?

      IR lidar would be a poor choice in a hot environment, and visible light lidar would have issues with smoke.

      The best choice currently, to map structures (and bare materials) through smoke, flames and steam, is probably radar using frequencies between 79 GHz and 300 GHz ( ref: doi:10.3390/s21020439 ). Higher frequencies are attenuated more at higher temperatures than lower frequencies, but provide higher resolution scans. Multiple scans at multiple frequencies would provide the best overall result in such a harsh environment.

      1. Trigonoceps occipitalis

        Re: Approach

        The best choice currently, to map structures is an Amazon Roomba.


  2. John Smith 19 Gold badge

    It's a Chinese system

    They State will naturally require you to supply such data to their central Fire Management Centre.

    Failure to do so will be "Unfortunate."

    As an actual deployable tool by F&R teams it's just a crap architecture. You cannot expect that level of inforamtion to be available or on site at a major incident.

    Probably the best you can get would be a data connector on the fire panel that would dump the data in a standard format to a F&R laptop running the model (noting which sensors had already been burnt out).

  3. Will Godfrey Silver badge
    Thumb Up


    I don't think it's viable currently, but I hope they continue working on it. It's the old saying:

    You never know what you can to until you try.

  4. AndrueC Silver badge

    And all that without excessive cranial hair. Impressive.

  5. Richard 12 Silver badge

    So it needs sensors around the building

    Those won't exist in domestic properties, probably ever.

    And in commercial buildings, the sensors aren't where the plans say they are. While many of them will be in the right room, very few will be in the right place.

    It's not like occupancy sensors, people generally don't notice if a fire detector is misaddressed - because the alarm still sounds.

    It's probably feasible to harden them enough to report the first hour, though less convinced by sufficient sensing range to not hit the endstop almost instantly.

POST COMMENT House rules

Not a member of The Register? Create a new account here.

  • Enter your comment

  • Add an icon

Anonymous cowards cannot choose their icon

Other stories you might like