Deep Learning in Science and Engineering
- Texas A&M University
- College Station, TX
- Joe C. Richardson Petroleum Engineering Building (RICH) 910
- Guni Sharon, Department of Computer Science & Engineering
- Learning the Right Tolls - Traffic Flow Optimization Through Micro-Tolling
Advancements in connected and automated vehicle technology present many opportunities for highly optimized traffic management mechanisms. One such mechanism, micro-tolling, has been the focus of a line of recently presented studies. In the micro-tolling paradigm, a centralized system manager sets a different toll value for each link in a given traffic network with the objective of optimizing the system's performance. Computing or learning this set of tolls (those optimizing the system's performance) is a challenging task unless making several restrictive assumptions. During this talk I will discuss the assumptions that enable computing such tolls. Next I will discuss more realistic settings where the restrictive assumptions do not hold. I will present black box optimization methods and reinforcement learning methods used to tackle such scenarios.