Joao Emilio Almeida

The problem of leaving a building while facing some emergency has gained much attention from the scientific community. Evacuation simulators can help emergency planners to assess the safety of buildings. However, such simulators still lack behavioural knowledge to increase their accuracy. The elderly have specific behaviour idiosyncrasies. This paper presents a set of experiments designed to elicit human behaviour in evacuation scenarios using serious games, applied to a population sample of 20 adults aged over 65 years. They were presented three experiments (alarm id, exit-choice in five different situations and auditorium with four situations) and their reactions were collected. Results were analysed and compared with previous similar experiments. Future directions in this research include: to expand and to refine data collection to other scenarios and perform massive data collection using the internet; and to use this data to feed and enhance existing crowd simulators particularly for evacuation purposes.

This paper focuses on the process of developing a framework that features Multi-Agent System and Fire Simulation in a game engine, resulting in a gaming infrastructure which may be played as a serious game to train users for emergency situations. Scenarios are generated dynamically with different types of characters controlled by the computer, Non-Playable-Characters (NPC). The computer controlled firefighters will go for rescue in an indoor fire scenario, while the player is able to either help them by playing a fellow firefighter or just watch the NPCs solve the situation. These NPCs are made using a Unity3D plug-in named RAIN to speed up the development process, which helps modelling the perceptions of the agents and their behaviours. These processes will be described and the result is a rescue simulation tool showing advantages in debugging behavioural models, which leverages the creation of these types of simulations. The less the player realizes that a firefighter is in fact an NPC being treated as another human player, the more likely the player is expected to behave socially. The current version of the simulation environment is available for download and free to play so readers can test and experiment with the implemented framework.

Understanding human behaviour in emergency evacuation from buildings is of utmost importance for fire safety designers, architects and engineers as they elaborate on strategies to improve the emergency paths to exits. This paper describes an experiment designed to elicit human behaviour when facing the urgent need of exiting a room of an unknown building. This test is part of a methodological approach that aims at the creation of a framework coined SPEED (Simulation of Pedestrians and Elicitation of their Emergent Dynamics). A population sample of 22 subjects was used to test such a methodological approach, which consists in having the subjects answering a questionnaire and later on, in playing a Serious Game. The game environment presents the same scenarios shown in the questionnaire using more elaborated 3D rendering to provide players with a sense of realism. The game was developed under the Unity3D game engine and based on the Serious Games concept. Preliminary results are promising, showing that the challenge made players think about the various situations that might happen when facing an emergency. They are also implied to reason on their stream of decisions, such as which direction to take considering the environment and some adverse situations, such as smoke, fire and people running on the opposite direction of the emergency signalling.

A discussion of the potential integration of serious games into the conceptual framework of artificial transportation systems focuses on behavior elicitation through peer-designed agents to model and simulate artificial societies on a participative basis.