Reliable Vehicle Routing Problem Using Traffic Sensors Augmented Information.

Area of Science:

• Transportation Engineering
• Network Science
• Data Science

Background:

• Stochastic routing faces challenges from uncertain travel times and limited sensor data.
• Traditional adaptive routing can be suboptimal due to dynamic traffic fluctuations.

Purpose of the Study:

• To develop a novel routing framework integrating traffic sensor data augmentation and deep learning.
• To improve the reliability of route selection and network observability in transportation networks.

Main Methods:

• Stochastic traffic assignment and multi-objective route generation.
• Optimal traffic sensor location selection.
• Deep learning-based traffic flow estimation using a Stacked Sparse Auto-Encoder (SAE).

Main Results:

• The framework determines minimum sensor deployment for accurate network-wide traffic estimation.
• The SAE model infers unobserved link flows, enhancing stochastic traffic condition observability.
• The data-driven approach significantly reduces sensor deployment needs while maintaining high traffic flow prediction accuracy.

Conclusions:

• The proposed framework offers a scalable and cost-effective solution for real-time traffic management.
• It addresses the gap between limited sensor availability and complete network observability.
• This enhances vehicle routing optimization in stochastic transportation networks.