Modeling the stochastic pedestrian fundamental diagram with explicit considerations of bidirectional conflict
Date:
Liang, H., Sun, J., & Ye, Y. (2023). Modeling the stochastic pedestrian fundamental diagram with explicit considerations of bidirectional conflict. The 27th Hong Kong Society for Transportation Studies International Conference, Hong Kong, China, December 11-12, 2023. (Poster Presentation)
Abstract: Pedestrian Fundamental Diagrams (PFDs) have been commonly used in model development and empirical analysis for pedestrian dynamics. However, traditional deterministic PFDs, which are usually calibrated using the regression-based method, fail to capture the stochasticity and heterogeneity of individual movement. Unlike vehicular fundamental diagrams (FDs), PFDs consider the intersecting effect as pedestrians usually move in a two-dimensional (2D) domain, which extends the calibration of the non-linear PFD model to a multidimensional problem. Therefore, this paper proposes a general probabilistic optimization model in the multidimensional space to calibrate stochastic PFDs based on scattered data samples, in which the unidirectional pedestrian flow and bidirectional pedestrian flow with oblique intersecting angle are considered separately. To numerically solve the multidimensional optimization problem an improved weighted least square method (WLSM) is developed. Finally, the proposed model and calibration framework are applied to a real-world dataset to validate its practicability.