University of Southampton
A study of the impact of path attributes on pedestrian route choice
Abstract
dc:description.abstractAs concerns over climate change continue to grow, reducing transport emissions has become increasingly important. Active travel, including walking and cycling, is therefore an important target for modal shift. However, pedestrian route information is still largely based on distance and often does not sufficiently reflect the walking environment or pedestrian preferences. This study therefore aimed to identify pedestrian route choice preferences and examine how they can be incorporated into pedestrian route. The study first analysed 48 discrete choice studies to examine the compatibility and heterogeneity of distance- and time-related effects. The subsequent pedestrian route choice analysis pooled distance- equivalent effects for six attribute subgroups with comparable definitions and measurement units. Substantial heterogeneity remained even after moderators were introduced, although several attributes showed relatively consistent effect directions. For the revealed preference analysis, GPS trajectories were processed and matched to an enhanced pedestrian network. A 30 m spatial threshold was adopted after examining the spatial effects of trajectory assignment, and a graph-based pre-processing method was developed for imperfect trajectory data. Filtering, clustering and graph-based methods were combined with Hidden Markov Model-based map matching, resulting in 2,981 observed trips. A common sample of 1,874 trips was used for comparison and increased to 2,057 observations after segmenting trips over 6 km. Path Size Logit (PSL) and Recursive Logit (RL) models were then estimated. The PSL analysis showed that coefficient estimates were sensitive to choice-set composition, although most coefficient directions were broadly consistent. The doubly stochastic method was selected after considering model fit, choice-set coverage and coefficient consistency. The RL model avoided explicit path generation but involved considerable numerical instability and computational difficulty. Finally, the estimated effects were converted into generalised distance and implemented within a pedestrian routing tool, showing that preference-based routes could differ from conventional shortest routes. A supervised link-level estimation method was also developed to improve route reconstruction. The results showed improved reconstruction performance compared with the Path Size Logit-based approach. Overall, the study demonstrates how pedestrian route choice analysis can be extended from preference estimation to routing, validation and route reconstruction.
Degree
thesis:*- Name dc:type.qualificationname
- Ph.D.
- Level dc:type.qualificationlevel
- doctoral
- Grantor dc:publisher.institution
- University of Southampton
- Year dc:date.issued
- 2026
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- Baek, Changyeob
- Advisors dc:contributor.advisor
-
- Blainey, Simon
- Kaparias, Ioannis
- Young, Marcus
- ORCID person:identifier.orcid
Rights
- Language dc:language
- English, eng
Identifiers
dc:identifier.*- DOI dc:identifier.doi
- https://doi.org/10.5258/SOTON/01/T008