Estimation of Route-Choice Behavior Along LRT Lines Using Inverse Reinforcement Learning

As the decline of public transportation in rural areas becomes a growing concern, initiatives to introduce attractive next-generation transportation systems to promote public transportation usage are being considered across various regions. In Toyama City, Toyama Prefecture, where the next-generatio...

Full description

Saved in:
Bibliographic Details
Main Authors: Tomohiro Okubo, Akihiro Kobayashi, Daisuke Kamisaka, Akinori Morimoto
Format: Article
Language:English
Published: MDPI AG 2024-12-01
Series:Inventions
Subjects:
Online Access:https://www.mdpi.com/2411-5134/9/6/118
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:As the decline of public transportation in rural areas becomes a growing concern, initiatives to introduce attractive next-generation transportation systems to promote public transportation usage are being considered across various regions. In Toyama City, Toyama Prefecture, where the next-generation light rail transit (LRT) system has been introduced, the number of users has significantly increased compared to before its introduction, with some users riding the LRT for the sake of the experience itself. On the other hand, there is a demand for a more micro-level and quantitative evaluation of the impact that the LRT has on the liveliness of areas along its route. Therefore, this study uses inverse reinforcement learning (IRL), a type of machine learning, to build a model that estimates route-choice behavior along the LRT lines based on behavioral trajectories generated from smartphone location data. The model is capable of evaluating the characteristics of location data with high accuracy. The findings indicate that routes along the LRT lines tend to be selected, suggesting that both the appeal of the LRT itself and the attractiveness of the spaces along its route contribute to this tendency.
ISSN:2411-5134