A Cognitive Human Behaviour Model for Pedestrian Behaviour Simulation

Claudia Hollmann

2015

Abstract

Pedestrian behaviour simulation models are being developed with the intention to simulate human behaviour in various environments in non-emergency and emergency situations. These models are applied with the objective to understand the underlying causes and dynamics of pedestrian behaviour and how the environment or the environment’s intrinsic procedures can be adjusted in order to provide an improvement of human comfort and safety. In order to realistically model pedestrian behaviour in complex environments, the specific human behaviour patterns which govern their behaviour need to be represented. It is thereby of importance to understand the causal chains between the surrounding conditions and the pedestrians’ behaviours: The impact of the environment’s purpose and facilities as well as the pedestrians’ individual goals on the pedestrians’ planning and route choice behaviour; the influence of emergent stimuli on the pedestrians’ plans and environment usage; the influence of the pedestrians’ environment usage under normal usage conditions on the pedestrians’ behaviour in response to a potential alarm event. In this thesis, a framework is developed for modelling advanced individual pedestrian behaviours and especially purpose-driven environment usage. The framework thereby aims to assist building and facility planners in improving a building’s layout in terms of pedestrian experience and walking routes.

In this thesis, a comprehensive review on how individual pedestrian behaviour and the pedestrians’ environment usage are realised in current pedestrian behaviour simulation models has been undertaken. In addition, current theories on human decision making, goal-driven behaviour and emotion modelling have been surveyed from the research fields of artificial intelligence, virtual reality simulation, human psychology and human behavioural sciences. From this survey, theories suitable for this thesis’ cause have been identified and combined for the proposed Cognitive Pedestrian Agent Framework (CPAF). The proposed framework contains a sophisticated human decision making model, a multi-faceted individual knowledge representation, a model to realise situational and contextual awareness, and a novel realisation of a human path planning heuristic. The proposed framework has been demonstrated in the simulation of a building usage-cycle use case. Further, it has been outlined how the proposed framework could be used to model experiential alarm response behaviour.
.