Study of car-bicycle safety at signalized intersections from multi-aspects

Cycling is increasing in popularity in urban areas due to its individual, social and environmental benefits. However, cyclists are among the most vulnerable road users. Especially at intersections, many bicycles collide with passenger cars despite the control of traffic signals. This thesis focuses on car-bicycle safety at signalized intersections. Various factors contribute to the occurrence of road accidents, namely factors from the human, vehicle and environment aspects. The aim of this thesis is to explore potential measures to reduce car-bicycle accidents at signalized intersection from multi-aspects. In accident prevention, one must know the world of accidents. Given that few accident analysis has specifically focused on car-bicycle accidents at signalized intersections, this thesis analyze two accident databases. The analysis deals with the characteristics of car-bicycle accidents at signalized intersections. It reveals possible accident scenarios, frequencies of each scenario and common accident causes of accidents. Naturalistic Driving Observation (NDO) is a fast-growing method for traffic safety studies. Despite of many NDO studies, few have particularly investigated interactions between car drivers and bicyclists at signalized intersections. This thesis carries out a Quasi-NDO study in order to investigate the interactions between car drivers and cyclists at signalized intersections from perspective of car drivers. A car is instrumented with various sensors. The instrumented car is used to collect the data of driving behaviors and environment, while twenty-two participants separately drive this car in real traffic. The collected driving behaviors include dynamic driving data, drivers´ body movements and eye movements. A self-programmed Graphical User Interface and a video annotation tool are used for data analysis in order to detect car-bicycle conflicts. In addition, the collected eye movement data is analyzed in the scenario (i.e. right-hook scenario), where car turns right and bicycle goes through. With 146 detected right-hook events, the analysis reveals bicycle-scanning strategies of car drivers. According to the bicycle-scanning strategies, potential suggestions are proposed to mitigate the risk of this scenario. Road users’ perceived risk influences individual behaviors, and therefore it plays an important role in traffic safety. Through an online survey, the perceived risk of car drivers and cyclists (no matter the consequences of the crash) are investigated for seventeen common car-bicycle scenarios at signalized intersections. A comparison between the subjective perceived risk und the objective risk shows that a discrepancy exists among both car drivers and cyclists. Moreover, it shows that cyclists tend to perceive less risk than car drivers do. The implications of these results are vital for improvement of car-bicycle safety at signalized intersections. Safety effects of bicycle facilities are controversial, especially at intersections. By use of the Negative Binomial model, safety effects of bicycle facilities on car-bicycle crash risk at signalized intersections are estimated. The effects of other intersection factors are simultaneously considered. The estimation results shows that bicycle lanes have positive safety effects, while bicycle paths have negative safety effects.

Author dc:creator

• Liu, Peipei

Advisor dc:contributor.advisor

• Marker, Stefanie

Licence dc:rights.uri

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

Language dc:language.iso

en

Identifier URI

http://dx.doi.org/10.14279/depositonce-7849

OAI identifier oai:identifier

oai:depositonce.tu-berlin.de:11303/8720