Learning Drivers’ Utility Functions in a Coordinated Freight Routing System Based on Drivers’ Actions

This project explores the probability of developing a centrally coordinated routing system for trucks. Such a system works in a way that prioritizes the maximizing of the system instead of lowering the individual cost, and balances the latter to be close or less than the user equilibrium cost with the use of monetary incentives. To study the potential of having such a system, it is vital to take into account what truck drivers actually do. Therefore, the project develops a method for estimating the utility functions of truck drivers based on their response to a centrally coordinated routing system that they are participating. Additionally, the project evaluates the impact of assumed wrong utility functions that could be viewed as non-compliance on the system optimum cost. To provide a context, centrally coordinated routing system for trucks will provide routing instructions based on optimizing the system rather than minimizing the individual cost. This distinction will allow for the efficient balancing of the individual cost. To reach the above-mentioned goals (developing a method and evaluating the impact), the project uses: (1) synthetic data and available testbed for a good control over the ongoing traffics, and (2) sensitivity analysis for testing the accuracy impact. 

Research Area

Tags