Pedestrian Modeling

Pedestrian movement is woven into the fabric of urban regions. From private indoor to public outdoor spaces, pedestrians are constantly utilizing their surroundings to reach destinations, explore their environment, and achieve specific goals. With more people living in cities than ever before, there is an increased need to understand and model how pedestrians utilize and move through space for a variety of applications, ranging from urban planning and architecture to security. 

We believe, agent-based models serve as an ideal tool to further such understanding, as agent-based simulations serve as artificial laboratories where we can test ideas and hypotheses about phenomena that are not easy to explore in the ‘real world’. One example is that of evacuation. For instance, without actually setting a building on fire we cannot easily identify people’s reactions to such an event. ABM, as with simulations in general, can allow for such experiments. Rather than setting a building on fire, we can re-create the building within an artificial world, populate it with artificial people, start a fire and watch what happens. Such simulations allow us to identify potential problems such as bottlenecks and allow for the testing of numerous scenarios such as the way various room configurations can impact on evacuation time. Such as in the first movie below we could explore what happens if we increase or decrease the size of the exit. By building such models we can focus on mitigation and preparedness rather than response and recovery in emergency incident management.

Building the spatial environment of such models is relatively straightforward, as we show in Figure 1, for example. From an architect’s CAD file of a building we can georeference the building to its actual ‘real-world’ location. We then take the building and rasterize the space so that each cell represents, say, 50cm2, which approximates the anthropomorphic dimensions of an individual. This regular lattice structure is used for our artificial world in a similar way to that of other pedestrian models; however, a continuous (i.e. vector) space representation could also be used if so desired (e.g. here). Once we have the spatial layout of the building, we can then populate the building with agents that have simple rules; for example, once the alarm is activated, agents follow emergency signage (or move to the nearest exit). This process is shown in Figure 1 (c) and (d). The simulation in the movie below demonstrates how bottlenecks form at exits and how this causes agents to cluster around such obstacles.

Figure 1: Simple pedestrian model: (a) CAD floor plans of a building are converted into (b) a raster layer and are used as the environment for the agent-based model. (c) shows the simulation running with agents (red) who are exiting the building and leaving behind walking traces (yellow). (d) shows a time series plot of agents evacuating from the building

If you want to run or download the models above, you can by clicking here and downloading the sample models for GeoMason. However, it is not just evacuation one can model but also how people use outside spaces for example. In the movie below we show how people might walk into a venue from their cars and how barriers (street furniture) can cause bottlenecks to form.

More recently we have been exploring how new data can help inform our agent-based models. Specifically, pedestrian modeling has been traditionally faced with the challenge of collecting data to calibrate and validate such models of pedestrian movement. With the increased availability of mobility datasets from video surveillance and enhanced geolocation capabilities in consumer mobile devices we are now presented with the opportunity to change the way we build pedestrian models. In the model below we explore the potential that such information offers for the improvement of agent-based pedestrian models.

For more information (and code) click here.

Selected Papers:
Crooks, A.T. (2015), Agent-based Models and Geographical Information Systems, in Brunsdon, C. and Singleton, A. (eds.), Geocomputation: A Practical Primer, Sage, London, UK, pp. 63-77. (pdf)
 Crooks, A.T., Croitoru, A., Lu, X., Wise, S., Irvine, J. and Stefanidis, A. (2015),  Walk this Way: Improving Pedestrian Agent-Based Models through Scene Activity AnalysisISPRS International Journal of Geo-Information, 4(3): 1627-1656. (pdf)
Crooks, A. T., Hudson-Smith, A., and Dearden, J. (2009), Agent Street: An Environment for Exploring Agent-Based Models in Second Life, Journal of Artificial Societies and Social Simulation, 12 (4): 10. Available at

No comments: