Su, Jingran (The University of Tokyo), Hato, Eiji (The University of Tokyo)

3D Indoor Pedestrian Route Choice Modeling Using Machine Learning-Based Wi-Fi Fingerprint Observations

Scheduled for presentation during the Regular Session "S05b-Deployment, Modeling, and Optimization in Intelligent Transportation Systems" (WE-EA-T5), Wednesday, November 19, 2025, 14:10−14:30, Surfers Paradise 2

2025 IEEE 28th International Conference on Intelligent Transportation Systems (ITSC), November 18-21, 2025, Gold Coast, Australia

This information is tentative and subject to change. Compiled on October 19, 2025

Abstract

Route choice modeling is an essential yet challenging problem in transportation research. Recently, increasing attention has been devoted to three-dimensional (3D) indoor pedestrian behavior analysis, driven by its importance for urban design, smart building operations, and emergency evacuation management. However, traditional GPS-based methods are unreliable indoors due to signal degradation and difficulty in capturing vertical movements, leading to the growing use of alternative localization signals such as Wi-Fi fingerprint data. Another issue lies in reconstructing accurate pedestrian trajectories for route choice analysis based on these signals, which are often sparse and noisy. In this study, we develop a comprehensive framework integrating ML-based link-level observation models, a temporal greedy map matching algorithm, and a weighted recursive logit model. The proposed approach enables robust 3D route reconstruction and extracts interpretable behavioral parameters, advancing pedestrian mobility analysis in complex indoor environments.