Dynamic guidance tool for a safer earthquake pedestrian evacuation in urban systems

In earthquake disasters, the leading causes of death are directly related both to build collapses and fatalities during the following evacuation phase. Allowing people to autonomously gain safe areas and assembly points should be the basis for reducing human losses in urban systems. However, some important environmental and behavioural factors (e.g. vulnerability of buildings, compact urban fabric, cascade effects, presence of people unfamiliar with the urban layout, absence of information on evacuation paths) can hinder this 'self-help'-based evacuation process. This issue is really important in historical centres where evacuees suffer a combination of unfavourable conditions to safely escape. This paper concerns a non-invasive solution for guiding people along probable safe evacuation routes in earthquake emergency. The proposed Seismic Pedestrians' Evacuation Dynamic Guidance Expert System (SpeedGuides) considers the influence of the main environmental and behavioural safety factors for evacuees (i.e. street vulnerability, street blockages probability, crowding conditions along paths, presence of mortal dangers, visibility conditions) and combines them in a safety index through the Multi-criteria techniques application. SpeedGuides dynamically collects safety factor data during the time and suggests the possible safest path to the nearest secure zone according to the Dijkstra's algorithm approach. SpeedGuides is an easy-to-use model proposed for application on personal devices (e.g. smartphone) that, taking advantage of different expert methods, allows evacuees to simple enhance their safety. A first effectiveness evaluation of SpeedGuides is provided through an earthquake pedestrians' evacuation simulator in a significant case study. The evacuee performances (with and without the proposed guidance tool) are compared and discussed. Results demonstrate how individuals' safety levels are increased when evacuees use SpeedGuides. (C) 2017 Elsevier Ltd. All rights reserved.