Five metrics analyzed, including evacuation start time and movement routes

Patent filed in July; proof-of-concept planned at workplaces in October

Samsung Fire [Samsung Fire]
Samsung Fire [Samsung Fire]

Samsung Fire & Marine Insurance announced Monday that it has developed an AI-based evacuation drill quantitative assessment system that uses artificial intelligence to analyze employee behavior during evacuation drills and evaluate training effectiveness and vulnerability factors.

The system links closed-circuit television footage with building spatial data to analyze, in quantitative terms, who began evacuating and when, which routes occupants took, and where bottlenecks formed during an actual drill. That sets it apart from conventional evacuation drills, which have centered on whether a drill was conducted and on-site inspections.

South Korean companies are required under the Act on Fire Prevention and Safety Management and related internal standards to conduct regular evacuation drills for employees. However, objectively measuring how long it took for evacuation to begin, whether people crowded specific routes or emergency exits, and how to systematically accumulate drill results and improvement points has proved difficult.

The system assesses evacuation behavior across five metrics: evacuation start, movement routes, bottlenecks, emergency exit usage and response spread. It measures the time elapsed between a fire alarm and the actual start of evacuation, analyzes the routes taken by occupants, identifies congestion at specific points, and detects uneven usage of individual emergency exits. The system also tracks how quickly one person's evacuation behavior prompts others to follow, helping identify behavioral patterns that could emerge in a real emergency.

Samsung Fire filed a sole patent in July for an "evacuation assessment method and device" related to the technology — covering a method for quantitatively analyzing and evaluating evacuation behavior using CCTV footage and spatial data.

"We expect this system will allow us to analyze evacuation drill results more objectively and identify vulnerability factors specific to each workplace," an official at Samsung Fire's Corporate Safety Research Institute said. "We will continue to improve its applicability in the field, expand the use of effective evacuation drill frameworks, and contribute to raising overall safety standards across society."

Samsung Fire plans to conduct a proof-of-concept at an actual workplace evacuation drill in October to verify the system's analytical accuracy and field applicability. The company said it will expand the system's scope of use based on the results.

Earlier, in May, Samsung Fire's Corporate Safety Research Institute unveiled AIMS, a real-time anomaly monitoring system co-developed with a power Internet of Things company. The solution monitors electrical equipment at industrial sites around the clock for signs of trouble such as current leakage, with the aim of preventing fires and equipment accidents before they occur. The move reflects a broader shift by fire insurers from compensating losses after an accident to actively reducing incidents and managing loss ratios.


won@heraldcorp.com