A System for Generating Customized Pleasant Pedestrian Routes Based on OpenStreetMap Data

Abstract

In this work, we present a system that generates customized pedestrian routes entirely based on data from OpenStreetMap (OSM). The system enables users to define to what extent they would like the route to have green areas (e.g., parks, squares, trees), social places (e.g., cafes, restaurants, shops) and quieter streets (i.e., with less road traffic). We present how the greenness, sociability, and quietness factors are defined and extracted from OSM as well as how they are integrated into a routing cost function. We intrinsically evaluate customized routes from one-thousand trips, i.e., origin⁻destination pairs, and observe that these are, in general, as we intended-slightly longer but significantly more social, greener, and quieter than the respective shortest routes. Based on a survey taken by 156 individuals, we also evaluate the system's usefulness, usability, controlability, and transparency. The majority of the survey participants agree that the system is useful and easy to use and that it gives them the feeling of being in control regarding the extraction of routes in accordance with their greenness, sociability, and quietness preferences. The survey also provides valuable insights into users requirements and wishes regarding a tool for interactively generating customized pedestrian routes.

Keywords: human–computer interaction; pedestrian routing; volunteered geographic information.

Figure 1

Main components of the proposed pedestrian routing system. (a) Origin and destination addresses are defined by the user. (b) Slide-bars to set the strength of influence of the different factors in the extraction of the customized route. (c) The system’s screen displays the shortest route (in red) and the customized route (in purple) over the OpenStreetMap layer. (d) A quantitative comparison between the shortest and customized routes in regard to the four factors is provided. The greenness variable is the street segment’s length multiplied by the relative area of the green areas inside its viewshed (see Section 3.2).

Figure 2

Example of a street segment and a 50 meter buffer zone around it. OpenStreetMap features intersecting the buffer and containing at least one of the tags indicate third places (Table 1) and were considered for the measurement of the street segment’s sociability.

Figure 3

Example of the visible green areas inside the viewsheds of the 100 m radius from the four observation points.

Figure 4

Distribution of noise intensity levels for each OpenStreetMap highway type. The noise intensity data from each street segment was collected and provided by the city of Heidelberg (Germany).

Figure 5

Graph plots of the $p_i$ metric computed for two trips, shown in (a) and (b), and the eleven discrete weight pairs summing to 10, i.e., (10,0),…,(0,10). Each pair of weights were applied to the length (L) and to the greenness (G), sociability (S), or quietness (Q) factor separately.

Figure 6

Shortest and alternative walking routes generated for trips 1 and 2 (Figure 5), shown in (a) and (b) in Heidelberg (Germany). The alternative routes were generated by setting a weight of 2 to the length factor and 8 to the other three factors separately.