Path Planning with Information on Travel Time Reliability

The shortest path under a given on-time arrival probability is valuable to travelers wishing to save time and avoid traffic uncertainty. Conventional reliable shortest path methods often assume that travel time follows normal distribution, but substantial evidence shows that travel time in peak periods follows lognormal distribution. This paper proposes a reliable shortest path method under the assumption that travel time in peak periods follows lognormal distribution. Due to the character of bi-objective optimization, first-order stochastic dominance was conducted to compare the priority of a set of paths. Then, an improved RSPP-LC algorithm that simplifies the path-correcting procedure was constructed and applied to a small network in Shenzhen. The computational result indicates that RSPP-LC algorithm performs well in a small-scale network.