Towards Data-Informed Sub-Models for Pedestrian Microsimulation of Transportation Terminals

Pedestrian microsimulation software is used to model and evaluate the circulation of pedestrians in transportation terminals. It can be used to determine potential travel times and visualize congestion during the design phase so that change and optimization can be made. Model parameters can also easily be changed to simulate a variety of scenarios, including peak travel times, events, and emergency evacuations. These models can be augmented and extended through the use of Software Development Kits (SDKs), including the introduction of new behaviours via modifications to the movement algorithm. However, these extended sub-models require validation. This presentation uses a study of a transportation terminal to demonstrate the evolution of a Pedestrian Microsimulation Sub-Model for group behaviour, the importance of real-world validation data, and potential data collection methodologies. Multiple versions of the group forces model are demonstrated as implemented within the pedestrian microsimulation software MassMotion, starting from initial concepts through to full implementations of potential group movement algorithms. Preliminary data collection efforts in an intercity transportation terminal are discussed, including the benefits and drawbacks of extracting movement and behavioural data from prerecorded video. New modifications to open-source motion tracking software will be demonstrated and compared with LIDAR and manual notetaking methods. Finally, a path forward will be discussed for employing video data analysis to assist with calibration, verification, and validation of Pedestrian Microsimulation Sub-Models. With further development, the use of data-informed sub-models has the potential to include more accurate movement and behaviours in pedestrian modelling and allow practitioners and engineers to design buildings, pedestrian spaces, and transportation terminals with higher confidence in pedestrian flow results.