With the emergence of internet and smartphones, many people are connected online to share vehicles. A lot of data is generated in this process which is used to improve real-time ride-sharing managed by companies such as Uber and Lyft. This has helped with the evolution of a more connected and centrally-controlled transportation structure and innovative systems for carpooling. By improving vehicle routing efficiency, there are economic improvements as well as a reduction of pollution. In taxi routing, or ride-sharing, it is decided which vehicle is allotted to a ride request. Since operations are online, decisions have to be online and prior information on demand is not available. These operations are centrally-controlled and on a large scale. The article presents a tractable rolling-horizon optimization strategy for online taxi routing that can be used in many similar situations. The formulation is based on a high degree of central control and with availability of prior information in ride-sharing systems. This approach can handle innumerable vehicles and many customers per hour and is applied to real taxi demand data in New York City.
展开▼
