摘要: |
New techniques offer the potential for improved schedule and route guidance for motorists, freight carriers and parcel delivery firms, by considering the variability of road travel times. This project will confront challenges to the implementation of these techniques, and demonstrate their feasibility and benefits using real data from the Chicago metropolitan area, one of largest transportation hubs in the US. Conceptually, the most reliable a priori routes can be found by solving the Dynamic Shortest Path problem with On-Time arrival reliability (DSPOT). DSPOT has recently been formulated using dynamic programming and solved by a non-polynomial algorithm (Nie and Wu 2007). The proposed research addresses two important issues that currently preclude its implementation: (1) development of solution algorithms fast enough for on-line application; and (2) validation using real data. In this project, historical traffic data from the Gary-Chicago-Milwaukee (GCM) traveler information system will be used to prepare dynamic probability mass functions of travel times, which are the key inputs to DSPOT. Then a prototype route guidance tool will be developed to implement DSPOT based on GCM data. This tool will be made available to the public through the Artificial Intelligence Laboratory at the University of Illinois at Chicago as an add-on toolkit to the GCM website. The ultimate goal of this project is to provide motorists and carriers with commercialized DSPOT products that will allow them to make tradeoffs between reliability and other costs and constraints. With the benefits and market value demonstrated through this project and further implementation stages, the project believes that software firms will be interested in adding DSPOT to their product offerings. These firms include but are not limited to the manufacturers of in-vehicle navigation systems, web companies that provide internet-based driving directions, and software vendors that produce logistics tools for freight carriers. |