Automated Proactive Road Safety Analysis: Towards Vision Zero in Toronto
Author(s): St-Aubin
Slidedeck Presentation Only:
Abstract:
Background/Context: As part of it's new Vision Zero road safety plan, the City of Toronto is exploring new technologies and approaches to achieve it's stated goal of eliminating fatalities and serious injuries on public roads. Specifically, the design philosophy of Vision Zero aims to design roads which account for human error present among road users, including pedestrians, cyclists, and motorists, in their day to day interactions on the road. The first step in this process is diagnosing specific issues and determining the effectiveness of existing infrastructure and potential safety countermeasures.
Aims/Objectives: Conflict analysis is a particular field of road safety analysis that examines these day-to-day interactions under a microscope, measuring how various elements of road design or changing driving conditions affect road user behaviour and any emergent safety margins that are a result of these driving conditions. More importantly, conflict analysis is one of several subdisciplines of road safety analysis that can be performed proactively. That is to say, it can be performed without the direct observation of undesirable traffic collisions. This obviates the ethical concerns and statistical significance issues that come with traditional road safety studies based on accident history.
Methods/Targets: This research project benefits from access to the deployment of two continuously monitoring traffic cameras at two major intersections in the City of Toronto and the latest technologies in computer vision and surrogate safety analysis to automatically perform a continuous conflict analysis study there.
Results/Activities: This continuous analysis allows for the investigation of temporal trends in road safety on an hour-by-hour basis on a number of safety-related factors including: weather effects, road works, special events, congestion, street lighting, sun glare, and more. A movement-by-movement analysis is also included.
Discussion/Deliverables: While the specific results determined within this study are valuable in their own right, the real value of this automated proactive road safety monitoring system is its capability of identifying human error in anticipation of more serious events, and potentially even before underlying causes are identified.
Conclusions: The case study performed using these two deployments serves as a template for a comprehensive real-time traffic safety performance monitoring platform at any location with a traffic camera.