Issue |
ITM Web Conf.
Volume 70, 2025
2024 2nd International Conference on Data Science, Advanced Algorithm and Intelligent Computing (DAI 2024)
|
|
---|---|---|
Article Number | 01010 | |
Number of page(s) | 8 | |
Section | Traffic Prediction and Analysis | |
DOI | https://doi.org/10.1051/itmconf/20257001010 | |
Published online | 23 January 2025 |
The Application of Reinforcement Learning in Traffic Flow Prediction: Advantages, Problems, and Prospects
1 SWUFE-UD Institute of Data Science at SWUFE, Southwestern University of Finance and Economics, Chengdu, Sichuan, 611130, China
2 School of Information Science and Technology, ShanghaiTech University, Shanghai, 200120, China
3 Department of Civil and Environmental Engineering, University of Michigan, Ann Arbor, MI, 48109, USA
* Corresponding author: jimmylee@udel.edu
Traffic flow prediction (TFP) is an important topic in the fields of operation research and traffic engineering. It is dedicated to predicting the flow of people and vehicles in the transportation network within a specific time frame in the future. Accurate TFP has great significance for traffic management, urban planning, road design, and the development of intelligent transportation systems (ITS). This article summarizes three traditional methods of TFP: parameter-based prediction, shallow machine learning-based prediction, and deep learning (DL)-based prediction. However, traditional TFP methods only focus on predicting time series in traffic data, and it is difficult for these methods to capture the interdependent relationship between the spatial distribution of traffic across a network and the temporal evolution of traffic conditions at each location. sequences. How to fully extract the spatiotemporal correlation of traffic flow (SCTF) is an urgent problem that needs to be solved based on DL prediction models. Concurrently, as science and technology advance, a growing variety of academics are attempting to incorporate reinforcement learning (RL) into TFP. Experimental results show that it can reduce vehicle queuing time and average delay to a greater extent, and alleviate air pollution. The article summarizes the models of DL and RL in TFP, comprehensively compares the benefits and drawbacks of various approaches, and proposes a vision for existing problems and future development.
© The Authors, published by EDP Sciences, 2025
This is an Open Access article distributed under the terms of the Creative Commons Attribution License 4.0, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.
Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.
Initial download of the metrics may take a while.