ITSC 2025 Paper Abstract

Close

Paper WE-EA-T9.1

Ishii, Kenta (Hitachi, Ltd.), Fumiyama, So (Hitachi, Ltd.), Otsuka, Rieko (Hitachi, Ltd.,)

Unsupervised Travel Mode Identification with Sparse Location Data

Scheduled for presentation during the Regular Session "S09b-Optimization for Multimodal and On-Demand Urban Mobility Systems" (WE-EA-T9), Wednesday, November 19, 2025, 13:30−13:50, Coolangata 3

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

Keywords Integrated Traffic Management for Multi-modal Transport Networks, Transportation Optimization Techniques and Multi-modal Urban Mobility, AI, Machine Learning Techniques for Traffic Demand Forecasting

Abstract

Regional public transportation in Japan faces declining the travel demand and driver shortages, necessitating enhanced operational efficiency through understanding of mobility patterns. For this purpose, travel mode identification using location data is crucial. To analyze entire regions, utilizing widely available location data is essential, but this presents significant challenges: sparse observation intervals and high labeling costs. We propose an unsupervised travel mode identification method specifically designed for sparse location data collected at intervals exceeding one minute, significantly longer than the 1-5 second intervals required by previous approaches. Our method formulates the problem as a latent variable estimation framework, enabling simultaneous estimation of travel modes and behavioral model parameters. This approach integrates (1) an observation model evaluating location data likelihood based on travel time consistency, node sequence matching and transfer feasibility, and (2) a behavior model accounting for local travel preferences. Experiments using GPS data with average intervals of 115 seconds from Chikushi area in Japan demonstrate an overall accuracy of 83.6%, with particularly robust performance in the identification of cars (86.8%) and railways (86.7%). This approach enables transportation planners to leverage widely available sparse location data for improving transportation planning without costly data labeling efforts.

 

 

All Content © PaperCept, Inc.


This site is protected by copyright and trademark laws under US and International law.
All rights reserved. © 2002-2025 PaperCept, Inc.
Page generated 2025-10-19  16:53:01 PST  Terms of use