Related Experiment Video
Updated: Jul 17, 2025
14:55
Evaluating the Effect of Roadside Parking on a Dual-Direction Urban Street
Published on: January 20, 2023
3.4K
Cycle-level traffic conflict prediction at signalized intersections with LiDAR data and Bayesian deep learning
Peijie Wu1, Wei Wei2, Lai Zheng2
1School of Traffic & Transportation, Chongqing Jiaotong University, 66 Xuefu Avenue, Nan'an District, Chongqing 400074, China.
Accident; Analysis and Prevention
|August 31, 2023
Summary
This study introduces advanced Bayesian deep learning models to predict traffic conflicts at signalized intersections in real-time. The Bayesian-Multi-head Stacked-LSTM Encoder-Decoder model demonstrated superior performance in predicting conflict frequency and uncertainty.
Area of Science:
- Transportation Engineering
- Artificial Intelligence
- Road Safety
Background:
- Real-time traffic safety prediction is crucial for proactive road management.
- Signalized intersections are high-risk areas for traffic conflicts.
- Existing models often lack real-time prediction capabilities and uncertainty quantification.
Purpose of the Study:
- To develop and evaluate Bayesian deep learning models for real-time traffic conflict prediction at signalized intersections.
- To assess model performance in terms of reliability, transferability, sensitivity, and robustness.
- To identify the optimal model for accurate and reliable traffic conflict forecasting.
Main Methods:
- Utilized LiDAR data for traffic conflict extraction and analysis.
- Developed a framework involving data preprocessing, base deep learning models, and Bayesian deep learning models.
- Employed Long Short-Term Memory (LSTM) networks, including Bayesian variants like Bayesian-Standard LSTM, Bayesian-Hybrid-LSTM, Bayesian-Stacked-LSTM Encoder-Decoder, and Bayesian-Multi-head Stacked-LSTM Encoder-Decoder.
- Applied the models to predict conflict frequency at the signal cycle level using historical traffic data.
Main Results:
- All four developed Bayesian deep learning models successfully predicted traffic conflict frequency per cycle per lane with associated uncertainty.
- Bayesian encoder-decoder models outperformed Bayesian-Standard LSTM and Bayesian-Hybrid-LSTM across all evaluation metrics.
- The Bayesian-Multi-head Stacked-LSTM Encoder-Decoder model was identified as the optimal choice due to its high reliability, transferability, low sensitivity, and robustness.
Conclusions:
- Bayesian deep learning offers a powerful approach for real-time traffic safety prediction at signalized intersections.
- The proposed framework and the optimal Bayesian-Multi-head Stacked-LSTM Encoder-Decoder model provide a reliable tool for proactive road safety management.
- This data-driven approach enhances the understanding and prediction of traffic conflicts, paving the way for safer road infrastructure.
Related Concept Videos
01:17Design Example: Analyzing Capacity Contours for Flood Risk Assessment
67
Flood risk assessment involves careful planning and analysis to ensure the safety of communities near water retention structures. Capacity contours are a vital tool in this process, as they illustrate the potential spread of water at specific levels in a given area. In the context of building a bund across a small valley, these contours play a critical role in evaluating the safety of nearby residential areas.In this example, the bund is intended to store stormwater in the valley. The engineers...
67
01:17Design Example: Alignment of a Road Line Using GIS
66
The alignment of a road line using Geographic Information Systems (GIS) is a critical process in civil engineering, combining advanced technology with practical decision-making. This methodology begins with the collection of geospatial data, including information on land cover, geomorphology, drainage patterns, slope, and contour details. Such data is typically acquired through satellite imagery and GIS tools, offering a comprehensive understanding of the terrain.Once the data is gathered, it...
66