Developing a travel route planner accounting for traffic variability

Abstract

Current travel route planners incorporate physical distances and mean travel times when calculating optimal routes between two points but ignore the variability in the travel times. This factor proves important to drivers who favor consistent travel times over routes that yield high variability in travel time. These routes are usually out of the way and therefore, have a higher expected travel time. However, studies have shown that travel time stability has become equally as important to commuters as short expected travel times [1], indicating many commuters may prefer the longer but more predictable routes. In order to address this problem, the team has developed a web-based travel route planner that calculates multiple routes for users. Each route balances a trade-off between expected travel time and variance in travel time. Routes are calculated using a probabilistic model of travel times that incorporates historical traffic data for the Virginia section of the I-95 corridor. The travel planner uses a shortest-path optimization algorithm to calculate several alternative routes. For given starting and ending locations, the planner returns three possible routes along with travel time information about each route. Each calculated route is associated with a different mean/variance preference combination.