Mesoscopic Modelling of Pedestrian Movement Using Carma and Its Tools

Galpin, Vashti and Zoń, Natalia and Wilsdorf, Pia and Gilmore, Stephen (2018) Mesoscopic Modelling of Pedestrian Movement Using Carma and Its Tools. ACM Transactions on Modeling and Computer Simulation (TOMACS), 28 (2), 11:1-11:26. ISSN 1049-3301.

Full text not available from this repository.
Official URL:


In this article, we assess the suitability of the Carma (Collective Adaptive Resource-sharing Markovian Agents) modelling language for mesoscopic modelling of spatially distributed systems where the desired model lies between an individual-based (microscopic) spatial model and a population-based (macroscopic) spatial model. Our modelling approach is mesoscopic in nature because it does not model the movement of individuals as an agent-based simulation in two-dimensional space, nor does it make a continuous-space approximation of the density of a population of individuals using partial differential equations. The application that we consider is pedestrian movement along paths that are expressed as a directed graph. In the models presented, pedestrians move along path segments at rates that are determined by the presence of other pedestrians, and make their choice of the path segment to cross next at the intersections of paths. Information about the topology of the path network and the topography of the landscape can be expressed as separate functional and spatial aspects of the model by making use of Carma language constructs for representing space. We use simulation to study the impact on the system dynamics of changes to the topology of paths and show how Carma provides suitable modelling language constructs that make it straightforward to change the topology of the paths and other spatial aspects of the model without completely restructuring the Carma model. Our results indicate that it is difficult to predict the effect of changes to the network structure and that even small changes can have significant effects.

Item Type: Article
Additional Information: Special Issue on FORECAST