Bernardini, G., Quagliarini, E., 2021. Terrorist acts and pedestrians’ behaviours: First insights on
European contexts for evacuation modelling. Safety Science 143, 105405.
Terrorist acts in the Built Environment (BE) are characterized by interactions between the
attackers, the exposed pedestrians and the BE itself. The analysis of real evacuation events
by videotapes allows researchers to have a potentially unbiased data source to this aim.
This research provides one of the first and wide databases of pedestrians’ behaviours in
terrorist acts, focusing on different characteristics of the BE (e.g. indoor and outdoor;
obstacles presence; safety management) and of the terrorist acts (e.g. type of attack;
proximity between the pedestrians and the trigger).
Since the analysis on how people behave in emergency conditions due to terrorist acts is still
limited, efforts to provide reliable (qualitative and quantitative) databases on such issue is
urgently needed and should especially take advantages of real-world events.
Firstly, videotapes of real-world emergencies due to terrorist acts were collected in a
database and each of them was divided into specific “scenes”. Then, qualitative analyses
were performed to detect emergency behaviours and assess their activation frequency as
well as the BE conditions that could induce their activation. Finally, quantitative analyses on
motion speeds were carried out to assess the pedestrians’ instantaneous evacuation speeds
and the fundamental diagrams
“Scenes” with deleted frames, inadequate illuminance or excessive camera movements were
removed from the database. “Scenes” from fixed cameras were preferred for quantitative
analysis. When the framing quality of these scenes allowed continuously tracking individuals’
motion over time, manual tracking methods for quantitative analysis were applied. The
manual method for tracking pedestrians’ motion in evacuation conditions was chosen,
because of the videotape’s characteristics (such as the not uniform backgrounds or the
resolution of the images).
65% of the “scenes” refers to videotapes from mobile cameras . 59 pedestrians were
randomly selected between those who could be tracked for a significant time in the “scene”,
i.e. some seconds or more. Basics statistics were also provided on the average speeds of the
traced individuals over the analysis time.
The videotape analysis highlights that no direct action by the supposed terrorists was noticed
by the crowd. In fact, the crowd simultaneously started to evacuate after having seen the
motion of a restricted group of pedestrians, without perceiving any other sign about the real
presence of a risk (phantom panic). When the pedestrians’ density increases, a significant
reduction in motion speed is noticed.
Despite the limitations due to the current sample dimension, quantitative results also remark
how researchers and safety designers should be discouraged from using general-purpose
databases. The main differences among data could be referred to different factors related to
European contexts for evacuation modelling. Safety Science 143, 105405.
Terrorist acts in the Built Environment (BE) are characterized by interactions between the
attackers, the exposed pedestrians and the BE itself. The analysis of real evacuation events
by videotapes allows researchers to have a potentially unbiased data source to this aim.
This research provides one of the first and wide databases of pedestrians’ behaviours in
terrorist acts, focusing on different characteristics of the BE (e.g. indoor and outdoor;
obstacles presence; safety management) and of the terrorist acts (e.g. type of attack;
proximity between the pedestrians and the trigger).
Since the analysis on how people behave in emergency conditions due to terrorist acts is still
limited, efforts to provide reliable (qualitative and quantitative) databases on such issue is
urgently needed and should especially take advantages of real-world events.
Firstly, videotapes of real-world emergencies due to terrorist acts were collected in a
database and each of them was divided into specific “scenes”. Then, qualitative analyses
were performed to detect emergency behaviours and assess their activation frequency as
well as the BE conditions that could induce their activation. Finally, quantitative analyses on
motion speeds were carried out to assess the pedestrians’ instantaneous evacuation speeds
and the fundamental diagrams
“Scenes” with deleted frames, inadequate illuminance or excessive camera movements were
removed from the database. “Scenes” from fixed cameras were preferred for quantitative
analysis. When the framing quality of these scenes allowed continuously tracking individuals’
motion over time, manual tracking methods for quantitative analysis were applied. The
manual method for tracking pedestrians’ motion in evacuation conditions was chosen,
because of the videotape’s characteristics (such as the not uniform backgrounds or the
resolution of the images).
65% of the “scenes” refers to videotapes from mobile cameras . 59 pedestrians were
randomly selected between those who could be tracked for a significant time in the “scene”,
i.e. some seconds or more. Basics statistics were also provided on the average speeds of the
traced individuals over the analysis time.
The videotape analysis highlights that no direct action by the supposed terrorists was noticed
by the crowd. In fact, the crowd simultaneously started to evacuate after having seen the
motion of a restricted group of pedestrians, without perceiving any other sign about the real
presence of a risk (phantom panic). When the pedestrians’ density increases, a significant
reduction in motion speed is noticed.
Despite the limitations due to the current sample dimension, quantitative results also remark
how researchers and safety designers should be discouraged from using general-purpose
databases. The main differences among data could be referred to different factors related to
