Towards Realistic Pedestrian Route Planning *

Simeon Andreev, Julian Dibbelt, Martin Nöllenburg, Thomas Pajor, Dorothea Wagner, Simeon Andreev, Julian Dibbelt, Martin Nöllenburg, Thomas Pajor, Dorothea Wagner
unpublished
Pedestrian routing has its specific set of challenges, which are often neglected by state-of-the-art route planners. For instance, the lack of detailed sidewalk data and the inability to traverse plazas and parks in a natural way often leads to unappealing and suboptimal routes. In this work, we first propose to augment the network by generating sidewalks based on the street geometry and adding edges for routing over plazas and squares. Using this and further information, our query algorithm
more » ... mlessly handles node-to-node queries and queries whose origin or destination is an arbitrary location on a plaza or inside a park. Our experiments show that we are able to compute appealing pedestrian routes at negligible overhead over standard routing algorithms. 1 Introduction The computation of routes in street networks has received tremendous attention from the research community over the past decade, and for many applications efficient algorithms now exist; see [4] for a recent survey. The bulk of work, however, focuses on computing driving directions for cars. Other scenarios, such as computing routes for pedestrians, have been neglected or simply dismissed as a trivial matter of applying a different cost function. We argue that this naïve approach may lead to unnatural and suboptimal solutions. In fact, pedestrians utilize the street network quite differently from cars, which is often not captured by traditional approaches. For example to save distance, pedestrians are free to deviate from the streets, using the walkable area of public open spaces such as plazas and parks. On the other hand, crossing large avenues can be expensive (due to traffic), and it may be faster and safer to walk a small detour in order to use a nearby bridge or underpass. In this work, we address the unique challenges that come with computing pedestrian routes. In order to obtain as realistic routes as possible, we propose to first augment the underlying street network model, and then to apply a tailored routing algorithm on top of it. After setting some basic definitions (Section 2), we propose geometric approaches for automatically adding sidewalks, calculating realistic crossing penalties for major roads, and
fatcat:wa66galerreuhi2nlevb6osluy