Many lessons have been learned on building emergency evacuation, with perhaps the steepest advances made through detailed analyses of the large-scale evacuations involved in events such as 1993 and 2001 World Trade Center attacks. Such studies have identified several key features of crowd evacuation behaviors, including significant delays before evacuation and the few factors that affect evacuation routes. Also, crowd disorder and blocking were well observed in events such as the nightclub fires of Rhode Island in 2003 and of Bangkok in 2009. These features have been explained through psychological theories and models, and viewed as crucial determinants for the evacuees? survival in emergency evacuation. However, one critical gap is that the wealth of such psychology-related knowledge encapsulated in the diverse and complex array of theories, models and simulation has not been integrated into the current methods for emergency evacuation. Consequently, intuitions are mostly used regarding the potential consequences of providing one set of guidance versus another to evacuees, and psychological factors that affect crowd behaviors are not systematically considered. With recent technological innovations on fire detection, crowd communication and guidance in emergencies, it is important to redress the serious gap between the state-of-the-art knowledge and our ability to effectively guide crowds to safety.
In the proposed research, an innovative model of crowds will be developed to incorporate key psychological behaviors of evacuees, including initiation delay, way-finding, and disorder and blocking. It will describe how situation information (e.g., perceived hazard and guidance received) changes crowd behaviors in the collective or group sense (e.g., crowd flow delays, directions, and rates), and enable the prediction of crowd movement. An optimization problem will be established to evacuate as many people and as fast as possible while reducing relevant risks through appropriate guidance on crowds by using, for example, dynamic exit signs or audio announcements. To efficiently solve this time-critical problem, advanced optimization methods will be developed and synergistically integrated within a divide-and-conquer approach to generate effective guidance. The models and methods will then be used to generate virtual reality experiments from the first-person perspective to validate psychological behaviors of participants. In addition, a Wiki-based platform will be developed to test, validate, and enhance models and methods through open sourcing of various modules and sharing of lessons learned for broad participation and impact.
With the involvement of Wiki-based platform participants, the research will have broader impacts on education, academic research, and engineering use; and the applications of the models and methods to other similar problems, e.g., emergency management of high school or university events or emergency evacuation of cities or regions. Furthermore, the research will be thesis topics for participating Ph.D. and MS students and special projects for undergraduate students, educating the next generation of engineers, psychologists, and homeland security leaders. Special efforts will be made to recruit minority and female students, including but are not limited to our participations in UConn School of Engineering?s ?da Vinci? program designed for high school mathematics and science teachers and counselor, the ?Multiply Your Options? program designed for 8th grade middle school female students, and the ?Engineering 2000? program designed for high school juniors and seniors. Our goals are to establish sound theory and methods for innovative modeling and optimization of time-critical and high-stake events while attracting and educating students. Ultimately, the proposed research will benefit society by saving lives and reducing injuries through better designed or configured evacuation systems, and through optimized crowd guidance in real time.
