Information impacts on route choice and learning behavior in a congested network: An experimental approach

摘要

Every individual traveler makes route choices in an inherently uncertain environment,due to random disruptions to the traffic system such as incidents and bad weather, and randombehavior of his/her fellow travelers. The premise underlying the development of AdvancedTraveler Information Systems (ATIS) that better-informed travelers make better route choicesboth at the individual and system levels should not be taken as granted but rigorously tested. Thispaper studies two types of information, namely en route real-time information on the occurrenceof an incident and ex post information on foregone payoffs (FPs), i.e., travel times on non-chosenroutes. Data were collected from an interactive experiment, where human subjects made multiplerounds of route choices in a hypothetical network subject to random capacity reductions, andtravel times were determined by performance functions of route flows from the previous round.Preliminary results and bootstrap statistical tests are presented. It is indicated that en route real13time information increases the network’s travel time saving and reliability under the specificsetting of the experiment, yet FP information has the opposite impact. The most efficientinformation structure in terms of travel time saving is a combination of real-time information andno FP information. Furthermore, the availability of real-time information at downstream nodesencourages participants’ strategic behavior at the origin. Last but not least, FP information seemsto increase risk-seeking behavior, and it encourages route switching without real-timeinformation, but suppresses them with real-time information. These results are potentiallyvaluable for policy evaluations regarding further developments of ATIS.