Conflation of OpenStreetMap and Mobile Sports Tracking Data for Automatic Bicycle Routing

Abstract

This article investigates how workout trajectories from a mobile sports tracking application can be used to provide automatic route suggestions for bicyclists. We apply a Hidden Markov Model (HMM)-based method for matching cycling tracks to a "bicycle network" extracted from crowdsourced OpenStreetMap (OSM) data, and evaluate its effective differences in terms of optimal routing compared with a simple geometric point-to-curve method. OSM has quickly established itself as a popular resource for bicycle routing; however, its high-level of detail presents challenges for its applicability to popularity-based routing. We propose a solution where bikeways are prioritized in map-matching, achieving good performance; the HMM-based method matched correctly on average 94% of the route length. In addition, we show that the extremely biased nature of the trajectory dataset, which is typical of volunteered user-generated data, can be of high importance in terms of popularity-based routing. Most computed routes diverged depending on whether the number of users or number of tracks was used as an indicator of popularity, which may imply varying preferences among different types of cyclists. Revising the number of tracks by diversity of users to surmount local biases in the data had a more limited effect on routing.

Mitigating Bias in Big Data for Transportation

“…Several companies offer sports-oriented platforms that can record the same travel observation data on either a dedicated bicycle computer or a smartphone app (Kitchel and Riordan 2014;Garmin 2018). These data sets provide an opportunity to address shortcomings in transportation data collected by agencies, but the use of the equipment and apps (often expensive) creates a significant sampling (or demographic) bias (Bergman and Oksanen 2016b). A review of the app users' demographics shows the primary representation of male, young, and middle-aged segments of the population (Griffin and Jiao 2015a, b;Boss et al 2018).…”

Mitigating Bias in Big Data for Transportation

“…Several companies offer sports-oriented platforms that can record the same travel observation data on either a dedicated bicycle computer or a smartphone app (Garmin, 2018;Kitchel & Riordan, 2014). These datasets provide an opportunity to address shortcomings in transportation data collected by agencies, but the use of the equipment and apps (often expensive) create a significant sampling (or demographic) bias (Bergman & Oksanen, 2016a). Review of the app users' demographics shows primary representation of male, young, and middle-aged segments of the population (Boss, Nelson, Winters, & Ferster, 2018;Griffin & Jiao, 2015a, 2015b.…”

A Geometric Framework for Detection of Critical Points in a Trajectory Using Convex Hulls

“…The main reason that causes poor accuracy in reporting trajectory points is GPS signal as recorded by a receiver. However, in many cases, such location error at the processing phase can be solved by map matching methods [65]. The main principle behind map matching methods is to minimize the distance between the projected path on the map and the input trajectory [66][67][68].…”