Route guidance ranking procedures with human perception consideration for personalized public transport service

The use of smartphone applications (apps) to acquire real time and readily available journey planning information is becoming instinctive behavior by public transport (PT) users. Through the apps, a passenger not only seeks a path from origin to destination, but a satisfactory path that caters to the passenger’s preferences at the desired time of travel. Essentially, apps attempt to provide a means of personalized PT service. As the implications of the Covid-19 pandemic take form and infiltrate human and environmental interactions, passenger preference personalization will likely include avoiding risks of infection or contagious contact. The personal preferences are enabled by multiple attributes associated with alternative PT routes. For instance, preferences can be connected to attributes of time, cost, and convenience. This work establishes a personalized PT service, as an adjustment to current design frameworks, by integrating user app experience with operators’ data sources and operations modeling. The work proceeds to focus on its key component: the personalized route guidance methodology. In addition to using the existing shortest path or k-weighted shortest path method, this study develops a novel, lexicographical shortest path method, considering a just noticeable difference (JND). The method adopts lexicographical ordering to capture passenger preferences for different PT attributes following Ernst Weber’s law of human perception threshold. However, a direct application of Weber’s law violates the axiom of transitivity required for an implementable algorithm, and thus, a revised method is developed with proven algorithms for ranking different paths. The differences between the three route-guidance methods and the effects of the JND perception threshold on the order of the alternative PT routes are demonstrated with an example. The developments were examined in a case study by simulation on the Copenhagen PT network. The results show that using the JND method reduces the value/cost of the most important attributes. Identical robust results are attained when JND parameters are not specified and default values are used. The latter may apply for the future with a mixture of specified and default preference input values. Finally, the computation time indicates a favorable potential for real-life applications. It is believed that the consideration of human threshold perception will encourage decision makers to establish new criteria to comply with this.