ABSTRACT
A wide range of factors affects risk perception during urban cycling. In this research, we investigate a factor so far not addressed: the spatial configuration of the vista space; with the underlying reasoning that more complex or rapidly changing vista spaces are perceived as more dangerous. We present a method to quantify a vista space’s spatial configuration by generating highly precise isovists based on laser scans. We test our assumptions in a lab-based study, where participants rate the perceived risk for cycling in various image sequences displaying complex traffic situations. The tested traffic situations are selected from a project collecting citizens’ reports about urban cycling risks. Our findings support the hypothesis that the spatial configuration of the vista space is a highly relevant factor for risk perception which deserves further investigation. Further analyses imply that volunteered geographic information on subjective topics such as risk perception may be of limited accuracy, but is likely to be representative for a larger population.
Acknowledgments
We are greatly indebted to Andreas Lars Wachaja and the department of Prof. Wolfram Burgard for providing us with access to a LiDAR scanner and the technical support in its application. We thank Philip Gutjahr of Visionsbox for providing the Ricoh Theta S camera system.
Disclosure statement
No potential conflict of interest was reported by the authors.
Notes
3. In other words, it was not our aim to investigate the spatial properties of a VGI’s precise geographical indication: We assume that users reporting a location as dangerous are less concerned with the specific point marked on a digital map, but more with indicating the general severity of the situation as well as the cycling trajectory triggering their perception of risk.
5. We could not identify significant differences between these locations and those selected from the VGI dataset in regard to the ‘total danger of cycling’ estimated by the participants (see Participants and Procedure), or the distribution of the participants’ attention including the position of attentional peak across the consecutive images (see ). We thus decided to analyse all locations in one sample.
6. There were two noticeable exceptions. At Test Location 14, the attentional peak appeared in the second of seven recording spots. We assume that participants focused on a hazard differing from that referred to by the underlying VGI contribution. We thus excluded this location from further analyses. Test Location 17 consisted of a large number of recordings and two distinctive attentional peaks at Spot 7 ( and Spot 26 (
. A closer comparison of the recorded images implied that Spot 26 was more in line with the underlying VGI contribution, and we thus defined Spot 26 as the attentional peak for Test Location 17.