In the past decade, transportation planners worldwide have been incorporating shared space design elements as a way of creating pedestrian-friendly places. Streets incorporating shared-space principles tend to have reduced vehicle speeds and increased safety for vulnerable road users. In North American cities, a shared-space approach is rarely applied to non-motorized environments such as pedestrian malls, campuses, and parks. As cyclists and pedestrians travel at relatively slow speeds, there is an opportunity to provide safe infrastructure to both through non-motorized shared spaces. Yet, little empirical evidence exists concerning the risk of pedestrian-cyclist collisions in shared spaces. Existing surrogate methods are either difficult to automate or insufficient to describe interactions between vulnerable users. To address this research gap, a methodological framework is proposed based on the analysis of semi-automated pedestrian-cyclist interactions and the integration of surrogate methods in non-motorized shared space. Several proposed surrogate safety measures (SSMs) including cyclist speed, angle of approach, pedestrian density, and post-encroachment time are analysed to estimate the risk of pedestrian-cyclist interactions. The methodology is then applied to a case study on the McGill University campus in Montreal, Canada, where cyclists and pedestrians coexist. User trajectories are automatically extracted using a computer vision software to yield 2739 pedestrian-cyclist interactions for analysis. The derived SSMs demonstrate adequate levels of safety. For example, speed and pedestrian density are shown to be negatively correlated, while conflict rate and density are positively correlated. Statistical differences are shown between conflict types defined based on intersecting angle and road user configuratio. (Author/publisher)
Abstract