Advancing Pedestrian Models: A Comparative Review and Vision for the Future

Abstract

Problem, research strategy, and findings Pedestrian mobility is essential for creating sustainable, healthy, and equitable cities, yet pedestrian modeling remains underdeveloped compared with vehicle-centric approaches. To advance the state of the art, we critically review four available pedestrian modeling frameworks—urban network analysis (UNA), multi-agent transport simulation (MATSim), model of pedestrian demand (MoPeD), and spatial design network analysis (sDNA)—through the lens of the traditional four-step transportation modeling framework. We assess their methodological foundations, capabilities, practical applications, and limitations and outline future directions for improving modeling practice. UNA and sDNA offer high-resolution, trip-based network analyses; MATSim supports agent- and activity-based multimodal simulations; and MoPeD estimates grid-level pedestrian demand. Despite these strengths, several key gaps remain: Most models focus predominantly on utilitarian walking, neglect leisure and social activities, typically assume homogeneous pedestrian behavior by overlooking sociodemographic variations, face shortcomings with mode choice estimation, and are rarely applied in informal urban contexts. Furthermore, limited availability of pedestrian count data continues to constrain effective model calibration and validation.

Takeaway for practice We propose that researchers and planners adopt a human-centered, inclusive, and policy-aligned modeling agenda, emphasizing simple yet intuitive metrics that capture the full spectrum of walking benefits, supporting early-stage planning even in data-scarce contexts, fostering stronger collaboration with practitioners, and promoting a modular, adaptable modeling ecosystem. Ultimately, reorienting pedestrian models as flexible decision support tools—rather than narrowly focused forecasting instruments—can meaningfully support the development of more walkable cities.