review.txt

This project incorporated ideas from a variety of sources in the literature, from general crowd modeling reminders, to requirements and fully implemented models. A good overview is, \cite{almeidaCrowdSimulationModeling2013}, which makes they point that there are three points for simulating: to generate observable phenomena, to test theories, and to test design strategies. They note that people tend to follow the path of least resistance while moving. The stress of emergencies leads to herding and flocking behavior, to the point of missing optimal escape routes, or engaging in inefficient "arching" to get through an escape, which is one of our avenues of inquiry. Layouts matter too, as brought out by \cite{mirahmadiNovelAlgorithmRealtime2012}, which points out that real-time generation of floor plans that are realistic is vital for systems such as  games. The offer a realistic house floorplan algorithm. This project focused more on office evacuation systems than residential applications.
Moving on to fire-specific models, additional fire-specific requirements were discussed by \cite{kuligowskil}, which outlines a requirement for a comprehensive fire evacuation model design capturing human behavior. The effort here was to capture the various theories and "behavioral facts" suitable for embedding in modeling tools, and then argued for including behavioral models in evacuation models. Also, \cite{abmEvac} provides an overview of major evacuation factors and is helpful for developing Agent-based simulations such as this one. This includes a Fire Dynamics Simulator (FDS) employing computational fluid dynamics and Geographic Information System (GIS) for modeling human responses. This is far more realistic than NetLogo can support. The physical aspects are balanced by \cite{kneidl} emphasizes that behavioral aspects of crowds are  important, including behavior, locomotion, and navigation. These had been modeled previously. This research relied upon graph-based methods to plot the exit course, whereas this model uses values attached to patches to drive the movement decisions for the Agents. In both this and our model, the presence of other agents affects movement calculations.
Two final models of note were one that was also implemented in NetLogo, \cite{prioritEvac} PrioritEvac. This is an Agent-based model that explores social science effects in evacuation response. Agents maintain distinct priorities supporting granular investigation of reactions. This is also a NetLogo model, and was validated against the Station fire, where pyrotechnics ended the career of the rock band Great White.
Finally, one pertaining to public space, and not fires, was \cite{zhouSimulationPedestrianEvacuation2019} This study begins with thirty-six hours of video from a public square in Ningbo, China, which were used to develop a Large-Scale Public Place (LPS) evacuation model, building upon the Social Force Model (SFM). This was expressed in five strategies. Attention was paid to walking speed and diameter, and a the model was used to study the efficiency of evacuating the area. The idea of starting with the human movement and not a map of an enclosed space.