A Personalised Routing Algorithm

Abstract

This research paper addresses the question why the preferred route pedestrians take to get from A to B is not necessarily the shortest route. The assumption is tested that the concept of familiarity affects route choice decisions. The research aims to create a quantitative measure for familiarity, and incorporate this with Dijkstra’s Shortest Path algorithm efficiently. A Familiarity Index is built, using personal GPS trajectories of ten participants. The results show that on average people are willing to walk 15% longer to walk through familiar space, as opposed to walking the shortest route in distance. They also show that people do not do what they say they would do. Quantitative shape comparison measures, such as the Discrete Fréchet Distance, were used to explore the differences between routes in more depth. Participants were asked to draw on paper maps the routes they would take from origin to destination, as a method of validation for the use of the Familiarity Index as a prediction for route choice. The research concludes that it is possible to quantify and implement familiarity as an edge weight and that a simulation on the basis of this concept is accurate to a certain extent.