Preferences for travel time under risk and ambiguity: Implications in path selection and network equilibrium

- We cite the SUE paper, and highlight the difference from it. The main difference is that SUE model still assumes that the travel time is deterministic, while our paper considers the case when the travel time is uncertain.
- We explain why central authority needs to consider travelers risk and ambiguity attitudes and System Optimum is defined based on the perceived travel time.
- We add references including Guo et al. (2010), Wang et al. (2014) and OHare et al. (2015) in the paper and highlight the differences.
- We add worst-case property of POA in the paper, especially, explain that the bound is tight.

In this paper, we study the preferences for uncertain travel times in which probability distributions may not be fully characterized. In evaluating an uncertain travel time, we explicitly distinguish between risk, where the probability distribution is precisely known, and ambiguity, where it is not. In particular, we propose a new criterion called ambiguity-aware CARA travel time (ACT) for evaluating uncertain travel times under various attitudes of risk and ambiguity, which is a preference based on blending the Hurwicz criterion and Constant Absolute Risk Aversion (CARA). More importantly, we show that when the uncertain link travel times are independently distributed, finding the path that minimizes travel time under the ACT criterion is essentially a shortest path problem. We also study the implications on Network Equilibrium (NE) model where travelers on the traffic network are characterized by their knowledge of the network uncertainty as well as their risk and ambiguity attitudes under the ACT. We derive and analyze the existence and uniqueness of solutions under NE. Finally, we obtain the Price of Anarchy that characterizes the inefficiency of this new equilibrium. The computational study suggests that as uncertainty increases, the influence of selfishness on inefficiency diminishes.