Revealing challenges and opportunities in cycling safety through crowdsourcing

An increased modal share of bicycle traffic is a key mechanism to reduce emissions and solve traffic-related problems. However, a lack of comfort and (perceived) safety keeps people from using their bicycles more frequently. To improve comfort and safety in bicycle traffic, city planners need (a), an overview of accidents, near miss incidents, and bike routes, (b), an overview of the cycling infrastructure's maintenance state, and (c), a way to study the effect of potential changes in the cycling infrastructure before implementing them in the real life. However, these needs, which are also of interest to cyclists and policy makers, are currently not being met satisfactorily. In this thesis, we present three main contributions to solve these problems. First, we show SimRa, a crowdsourcing-based citizen science project to record cycling trips and near miss incidents. As part of this contribution, we also present CycleSense, an approach based on Deep Learning to automatically detect near miss incidents from recorded cycling trip data. Knowing the hazardous bicycle traffic hotspots, policy makers and city planners can approve and implement safety improvements for cyclists. Cyclists, on the other hand, can use this information to circumvent these spots. Second, we describe an approach for deriving the road surface quality from cycling trip data. Our approach uses data produced by the inertial measurement unit of the smartphone to calculate the smoothness of the road surface from the vibrations caused by traversing on the road with a bicycle. We then show how our approach can easily be integrated into a cycling trip recording app with SimRa as a case study. This contribution also contains a visualization of the cycling infrastructure's smoothness and helps policy makers, city planners and cyclists in the same way as our first contribution. Third, we improve the cyclist simulation model of the urban traffic simulation software SUMO. For this, we first show that the default cyclist model of SUMO is not realistic by comparing it to cyclist behavior in the SimRa dataset. We then split the cycling trips in the SimRa dataset into three parts for slow, medium, and fast cyclists and implement our findings as a plugin that can be added to SUMO for a more realistic cyclist simulation. This new cyclist simulation can provide a strong tool for city planners to better to study effects of potential changes in the traffic infrastructure.

Author dc:creator

• Karakaya, Ahmet-Serdar

Advisor dc:contributor.advisor

• Bermbach, David

Licence dc:rights.uri

• https://creativecommons.org/licenses/by/4.0/

Language dc:language.iso

en

Identifier URI

https://doi.org/10.14279/depositonce-22336

OAI identifier oai:identifier

oai:depositonce.tu-berlin.de:11303/23522