Multi - Criteria Route Planning with Emphasis on Geospatial Result Diversity

People travel all the time, whether it be commuting, weekend bike trips or transportation of goods. We often require our smartphones or computers to find the quickest route for us to get somewhere. But what if we are not interested merely in the shortest or the fastest option and, instead, desire to choose from several alternatives that meet our expectations in a whole range of criteria? We know that searches accommodating multiple preferences simultaneously often find an excessive number of results, the ultimate selection from which may be awkward. We therefore consider the options for reducing the solution set’s size. Our intention is to identify a few representative solutions that exhibit spatial diversity. To accomplish this, we design a multi-criteria algorithm that integrates the preference of geographically distinct solutions directly into the search, and we compare it with the method that simply filters the complete set of solutions based on their distinctness. We perform the evaluation on the road networks of the city of Prague and New York City. The proposed search algorithm indeed finds a set of routes with a convenient size, still covering all of the interesting route segments. The evaluation also shows that it is on average up to two times worse in terms of runtime, although there are certain cases in which it clearly outperforms the filtering method.