Scooter Riders at the Signalized Intersection: A Deep Learning Approach

@inproceedings{Causevic2025,
title = {Scooter Riders at the Signalized Intersection: A Deep Learning Approach},
author = {Azur Causevic and Jens Brauer and Fabian Pfeuffer and Klaus Bogenberger
},
editor = {Andras Kemeny and Jean-Rémy Chardonnet and Florent Colombet and Stéphane Espié},
doi = {https://doi.org/10.82157/dsa/2025/14},
isbn = {978-2-9573777-7-0},
year  = {2025},
date = {2025-09-24},
booktitle = {Proceedings of the Driving Simulation Conference 2025 Europe XR},
pages = {117-124},
address = {Stuttgart, Germany},
organization = {Driving Simulation Association},
abstract = {Due to factors such as urbanization and economic growth, global developments are showing an increasing demand for mobility in urban areas. Navigating one’s path in these areas is usually accompanied by flexible and dynamic decision making, taking thereby in consideration surrounding vehicles and traffic rules. This is especially emphasized for two-wheeled vulnerable road users (VRUs), which have a higher degree of movement freedom compared to heavier vehicles like cars and trucks. This poses challenges in predicting their trajectories and understanding their decision-making. Furthermore, assessing the behavior of VRUs is important in context of developing safety functions for newer vehicle models. As a requirement to that, trustworthy depiction of VRU-behavior in driving simulation is of utmost importance. We propose a variational auto-encoder model to sequentially predict the next 1 second of the movement/trajectory, based on the last captured 1 second of trajectory and the environmental variables. We train and test our model on a dataset of a 15-minute drone-recorded signalized intersection in Shanghai, China. Results suggest that we are able to implicitly capture the underlying trajectories and decision making for various movement patterns.
},
keywords = {},
}

Download .bib file

TY  - CONF
TI  - Scooter Riders at the Signalized Intersection: A Deep Learning Approach
AU  - Causevic, Azur
AU  - Brauer, Jens
AU  - Pfeuffer, Fabian
AU  - Bogenberger, Klaus
C1  - Stuttgart, Germany
C3  - Proceedings of the Driving Simulation Conference 2025 Europe XR
DA  - 2025/09/24
PY  - 2025
SP  - 117
EP  - 124
LA  - en-US
PB  - Driving Simulation Association
SN  - 978-2-9573777-7-0
L2  - https://proceedings.driving-simulation.org/proceeding/dsc-2025/scooter-riders-at-the-signalized-intersection-a-deep-learning-approach
ER  - 

Download .ris file