Edinburgh Napier University

  The paper discusses the need for an autonomous agent approach for the modeling of pedestrians in urban
environments and places PEDFLOW in the context of existing models. PEDFLOW is a microscopic model
of pedestrians’ movement, where each pedestrian is represented as an agent capable of making its own
decisions based upon a part of the observable scene local to that pedestrian. The model, implemented in
Java, provides a framework where agents are visualised as squares in a grid and movement is modelled as a
change of grid position with a delay that characterises the speed of the agent. A single rule set is utilised
that is made specific to each agent by the incorporation of parameters characterising ‘types’ of pedestrians.
The rules originate from computer aided analysis of video footage and are transformed into a form that can
be efficiently processed by the agent. By adding tools to extract measures of pedestrian flow, the
PEDFLOW model will be made useful to urban planners to evaluate infrastructural changes intended to
promote walking in the urban environment.