A conservative expected travel time approach for traffic information dissemination under uncertainty

Abstract

Travel time is important for both users and planners. Travel time variability widely exists in transportation systems. Positive skew with a long/fat upper tail is a basic characteristic of travel time variability, which poses great challenges to reporting travel time information to the public. There are two main approaches for disseminating travel time variability information: the traditional expected value and the percentile-based value. They are either unreliable when facing highly-skewed travel time variability, difficult to understand, or sensitive to individual risk preference. Motivated by this observation, this paper develops a conservative expected travel time (CET) approach to enhance the information reliability and maintain simplicity, which is formulated as a two-stage problem: (1) reshaping the travel time distribution, and (2) minimizing the expected value of the reshaped distribution. The features of the CET and the relationship between the CET and traditional measures are explored in both theoretical and numerical dimensions.