Analysis of the use of behavioral data from virtual reality for calibration of agent-based evacuation models

The Impact of Postures and Moving Directions in Fire Evacuation in a Low-Visibility Environment

“…This suggests that the effects of slight smoke may be equivalent to the dark situation applied in this study while under heavy smoke, the average speed soon decreased [ 41 , 45 ]. Meanwhile, the acquired results were also similar to that of Juřík et al [ 9 ], though the visibility conditions were different. This may be somehow related to the unfamiliarity to the environment of participants in Juřík et al [ 9 ] and it implies that the unfamiliarity can have relatively similar effects as visibility on walking speed when moving in a horizontal direction.…”

“…A comparison of this study with previous studies is presented in Table 5 (the studies have been grouped based on illumination level, i.e., 100% and ≤10% visibility). It can be seen that the works by Cao et al [ 27 , 28 , 32 ], Jeon et al [ 49 ] and Juřík et al [ 9 ] have similar experimental setups to this study except for their visibility factors and the additional factor of the volunteering participants’ familiarity of the building and its environment [ 12 ]. Thus, the following comparisons mainly focused on these works.…”

“…In previous studies [ 9 , 22 , 23 , 27 , 28 , 32 , 33 , 35 , 38 , 39 , 41 , 43 , 45 , 49 , 52 , 53 , 60 ] where upright walking posture was the main focus, the average horizontal speed of participants was shown to vary greatly as can be seen in Table 4 . In this study, measurements were carried out under different scenarios with focus on different postures and speeds.…”

“…The accuracy of simulations using such software becomes even more crucial as the data on the characteristics of occupants, their actions during evacuation, delays that may occur, and travel speeds for different types of occupants depicting evacuee behaviours become a critical requirement [ 6 ]. These data are relatively deficient with a limited amount of fire events and experiments [ 4 , 7 , 8 , 9 ]. According to Fahy and Proulx [ 6 ] and Ronchi [ 10 ], data collection for evacuation models can generally be divided into six categories, as follows: Pre-movement time, i.e., the time period between alarm triggering and the beginning of occupant movement.…”

“…In previous studies, the walking velocities using different postures were usually measured with a professional set of cameras and/or electromyography (EMG) equipment [ 27 , 28 , 34 , 35 , 48 ]. These methods either require more than ten fixed cameras at the test site [ 9 , 20 , 34 , 35 , 38 , 40 , 48 , 49 , 50 , 51 , 52 ], or a self-developed equipment set with attached cameras [ 27 , 28 , 32 , 33 , 53 ]. However, this kind of method may not be functional when the cameras are covered by the smoke of a fire event, and thus may only be used under experimental conditions without vision impediment.…”