Immersive real-time multi-user interaction with computer simulated pedestrians during emergencies

Crowd management is a key issue in large public assemblies in both emergency and nonemergency scenarios. It is important to be able to train for crowd management and test crowd management procedures for both emergency and non-emergency scenarios, however, full-scale exercises are costly, difficult and time consuming, and prevent exposure to immersive fire, smoke and toxic products. Interactive virtual training environments (VTE) with validated pedestrian dynamics are non-existent in firefighting and law enforcement training. Human behaviour is often made to appear real in VTEs rather than qualitatively correct and direct interaction with simulated agents is either not possible or limited to one-to-one interaction. An Immersive Real-time Interactive Simulation (IRIS) system is proposed to link a validated pedestrian behaviour computer model with an interactive virtual reality (VR) environment and Computational Fluid Dynamic Fire data produced by a CFD Engine. This 3-way link enables new use cases for using pedestrian behaviour models in new fields, for example real-time immersive and interactive first responder training with reactive crowd behaviour in fire scenarios. The linking of the software components also enables rapid prototyping, development, testing and verification of new behaviours and capabilities that are not currently present in the linked computer model. The IRIS system allows the model to be used for testing and prototyping of crowd procedures and theories in an interactive multi-user VR environment. To improve immersion in interactive VR fire scenarios, it is important to visualise smoke accurately and to represent the impact of heat and toxic products of the fire on the simulated population and user-controlled agents. A novel, real-time, accurate fire smoke visibility visualisation system is presented. Previously, accurate fire smoke visibility visualisation has been limited to expensive ray tracing algorithms as a post-processing tool, not allowing real-time virtual environments to benefit from accurate fire smoke visibility representation.

Name dc:type.qualificationname

phd

Level dc:type.qualificationlevel

doctoral

Grantor dc:publisher.institution

University of Greenwich

Year dc:date.issued

2018

Author dc:creator

• Haasanen, Simo Hannu

Advisors dc:contributor.advisor

• Lawrence, Peter
• Galea, Edwin
• Gwynne, Steve

• QA Mathematics

Language dc:language

en