A Deep Learning Approach for Lane-Based Short-Term Traffic Volume Prediction at Signalized Intersections

Considered the state of the art when it comes to signal control systems currently deployed in the field, adaptive traffic signal control is the focus of this paper. The next generation of traffic control systems in urban areas is considered to be Connected/Automated Vehicle (CAV)-based traffic control. For most of these control methods, traffic data/information at the current/past times can be collected and prediction models need to be applied to forecast short-term traffic conditions, based on which to adjust traffic control plans, with or without a physical signal at the intersection. An important measure for traffic signal control and optimization is lane-based traffic volume data. In order to overcome various limitations in predicting lane-based short-term traffic volumes, the authors develop an approach which considers both spatial and temporal features by the convolutional neural network (CNN) and long short-term memory (LSTM), respectively. This novel deep learning approach is used for lane-based short-term traffic volume prediction at signalized intersections; the details of this approach are discussed further in this paper. The authors evaluate the proposed model using simulation data, and in order to demonstrate the model's effectiveness, its performance is compared with several state-of-the-art models.

• Authors:
  • Li, Wan
  • Ban, Xuegang (Jeff)
• Conference:
  • Transportation Research Board 98th Annual Meeting
  • Location: Washington DC, United States
  • Date: 2019-1-13 to 2019-1-17
• Date: 2019

Language

• English

Media Info

• Media Type: Digital/other
• Features: Figures; References; Tables;
• Pagination: 6p

Subject/Index Terms

• TRT Terms: Adaptive control; Autonomous vehicles; Connected vehicles; Machine learning; Mathematical models; Mathematical prediction; Neural networks; Signalized intersections; Traffic signal control systems; Traffic volume
• Subject Areas: Highways; Operations and Traffic Management; Planning and Forecasting;

Filing Info

• Accession Number: 01716136
• Record Type: Publication
• Report/Paper Numbers: 19-05692
• Files: TRIS, TRB, ATRI
• Created Date: Sep 13 2019 12:19PM