DAY 2Track 3

Deep-learning-powered Fire Evacuation System for Subway Station

When a fire occurs inside a complex building or a subway station, guiding people to the safest exit by considering the location of the fire, the intensity of the fire and the distribution of people in real time has a significant impact on casualties. This presentation introduces an example of fire escape guidance system including deep-learning, IoT fire detector, and laser exit indicator. The model generated by deep-learning is used to evaluate the fire risk level inside the subway station using IoT sensory data and to determine the safest exit route. As a result of installing and operating the developed system at Daejeon City Hall Subway Station, it is proved that it can guide the passengers to safe evacuation route by reflecting various factors such as fire status in real time.



Hyung-Suk Han Principle Researcher, KIMM

Hyung-Suk is a principal researcher at KIMM(Korea Institute of Machinery & Materials)’s AI Machinery department, leading R & D that applies AI to a variety of machines. Hyung-Suk has been involved in the development of magnetic levitation trains, logistics robots and automobiles with expertise in computer simulation. He published an academic book titled “Magnetic Levitation” in 2106. He received Ph.D. in mechanical dynamics simulation in 1997 and developed a general-purpose dynamics simulation SW.