Edinburgh Napier University

  In networks with uncertain link travel times route guidance information should consider traveller preferences (1). Risk-averse travellers prefer safer options but might feel unsatisfied if it turns out that their decision has been based on wrong expectations. This paper therefore proposes a new approach to take regret, disappointment and risk aversion into consideration in route navigation. Limited information is assumed: The router knows minimum and maximum travel times of links but not the probability distribution of link travel times. Routes using only potentially optimal links are chosen by weighting the travel times scenarios expected on a given path and taking into account the cost of alternative options. Two approaches are explored, referred to as global and local search. The local search can be implemented as a fast method avoiding path search by guiding travellers link by link. As a drawback local search potentially can lead to cyclic paths. Possibilities to overcome the cyclic path issue and boundary properties of both approaches are discussed. Monte Carlo simulation on a US network suggests that the two approaches can provide good solutions to the problem of route choice under severe lack of information. The local search seems to be particularly promising.