Vulnerable road user safety and efficiency at intersections

This project leverages AI-driven video analytics to improve safety and efficiency for vulnerable road users (VRUs), including cyclists and pedestrians, at Melbourne intersections.

The project will deploy a safety index to identify risk factors and inform data-driven improvements by detecting collisions, near misses, and unsafe interactions. Integration with the Department of Transport and Planning (DTP) infrastructure will then enable intelligent prioritisation of cyclists and pedestrians at intersections, enhancing safety and efficiency while encouraging active transport.

These innovations aim to reduce carbon emissions, promote sustainable travel, and make the transport network safer and more inclusive for all road users.

Participants

• Department of Transport Victoria
• Cubic Transportation Systems
  
• ITS Australia
• University of Melbourne

Project background

This project explores how technology can help transportation agencies achieve their strategic goals through AI and computer vision to detect and optimise intersection behaviour for VRUs. This offers the possibility of real-time actions to improve outcomes and generates a richer set of data on which to base road infrastructure and multi-modal intersection planning. This includes real-time incident and near miss detection and road user behaviour analysis for all road users through video analytics.

A city’s strategic plan

Transportation agencies worldwide have a strategic goal focussed on the reduction if not the elimination of road deaths and serious injuries, but also to consider their Net Zero policy and citizen health. For VRUs, these objectives are interrelated, where cycling and walking safety is central to the uptake of active transportation, and aids in the reduction of single occupancy car use.

The perception of safety is important for a VRU, with a proven relationship between a greater sense of road safety and a corresponding increase in the use of active transportation modes.

Urban intersection safety

Intersections are key to urban road safety, with up to 50% of all accidents happening at or near an intersection. Collisions include single- and multi-vehicle crashes and crashes in which pedestrians, cyclists or other VRUs are hit while crossing the intersection.

DTP’s policy position

DTP has a policy position to increase active transport mode share to 25% of all trips by 2030. The role of video analytics to support these policy ambition needs to be researched locally to support these policy goals.

DTP, amongst other Australian transport agencies, is investigating the use of Cooperative Intelligent Transport Systems (C-ITS) to meet the ambitious goal of zero road deaths by 2050. The integration of C-ITS and video analytic platforms that can detect VRUs has the potential to significantly reduce the number of pedestrian/cyclist and motor vehicle incidents at intersections.

What is planned to be achieved or advanced?

The program will demonstrate how computer vision technology can be utilised to improve VRU safety and efficiency (congestion and emission reduction) at intersections and encourage active transportation uptake.

The initial focus will be on defining the required outcomes (prioritisation and awareness), categorising behaviour and VRU safety issues, and identifying the actions required to achieve those outcomes.

For the analysis, it is planned to use existing computer vision sensors (intelligent cameras by Cubic) installed in Melbourne.

Additionally, it is proposed to research and quantitatively determine how prioritising cyclists through video analytics can have a real and meaningful impact on their ability to traverse signalised intersections.

Project objectives

The project’s objectives are to:

• Identify appropriate responses to enhance VRU safety once pedestrians and/or cyclists are detected within a defined zone.
• Develop a repeatable and scalable integration between edge devices and SCATS, enabling the system to request signal priority or modify signal phasing to benefit VRUs.
• Measure real-world efficiency gains achieved through the intelligent prioritisation of cyclists at signalised junctions.
• Assess the extent to which imminent incidents can be detected and automatically resolved in real time, and identify suitable resolution actions to prevent incidents involving VRUs.
• Determine the factors that define an “unsafe junction” for VRUs, analyse how these factors impact safety, and evaluate how increased awareness and targeted actions—such as adjustments based on environmental conditions, time of day, intersection layout, cycle lane locations, and pedestrian crossings—can improve safety for VRUs.

Please note …

This page will be a living record of this project. As it matures, hits milestones, etc., we’ll continue to add information, links, images, interviews and more. Watch this space!