Moving safely through traffic chaos

James Elder Horst Herget

Toronto's traffic has been called the worst in North America, worse even than New York City and Los Angeles. As a pedestrian or cyclist, it is a tricky, sometimes deadly, maze to navigate. For someone with a disability, perhaps in a wheelchair, the challenges are compounded. Even robots have a tough time of it. The job of monitoring traffic flows, coordinating traffic signals, and ensuring the millions of vehicles and people on our city streets are moving freely and safely is highly complex.

Anyone stuck in a jam would agree, and as chunks of the Gardiner Expressway are removed or reconstructed the jam-ups in some spots seem to be getting worse.

Researchers at York University are working on various tools to make moving through and about cities easier, regardless of mode of transportation. “We need reliable, sustainable, fully automatic traffic analytics systems that continuously provide accurate traffic metrics,” says Professor James Elder, director of York’s Centre for AI & Society (CAIS), and a member of York’s Centre for Vision Research, Vision: Science to Applications and Connected Minds research programs.

He also led the Intelligent Systems for Sustainable Urban Mobility (ISSUM) project from 2017 to 2023, working with colleagues at York, the University of Waterloo and partners, including the Ontario Ministry of Transportation (MTO), Esri Canada, Trans-Plan, Peel Region, and York Region, with close to $4 million in funding through an Ontario Research Fund - Research Excellence award. This project has led to several new initiatives to translate foundational research into prototypes and commercial products that have real-world impact.

“We are researching and developing AI [artificial intelligence] technologies for better, real-time understanding of mobility in urban environments and metropolitan areas, for sensing, analysis, simulation and 3D visualization primarily using computer vision to understand traffic flow,” says Elder of York’s Lassonde School of Engineering and the Faculty of Health, and York Research Chair in Human and Computer Vision.

Technologies developed in Elder’s lab use video data to derive accurate 3D geopositioning and classification of road users, including cars, trucks, buses, pedestrians and cyclists. From the raw data, critical mobility intelligence is extracted in the moment, including traffic density, speed and volume, used to optimize traffic signaling and planning, and identify traffic incidents.

Traditionally, traffic cameras are hardwired to the internet to transmit high-bandwidth raw video data to central traffic offices. In contrast, Elder’s team has developed specialized computing technologies that process the video data “on the edge” so only derived anonymized mobility intelligence is transmitted. He explains that this has the dual benefit of only requiring inexpensive and flexible low-bandwidth cellular transmission and preserving the privacy of road users.

This edge-computing approach also allows flexible deployment of traffic analytics systems using temporary camera installations, and increasingly, drone platforms, which have a privileged bird’s-eye view of complex traffic interactions.

“Processing the data in real time allows us to understand traffic flows and disruptions as they’re happening,” says Elder. These disruptions are often the result of major construction projects or sporting events.

“If you're running a FIFA World Cup event, you need to know within seconds or minutes how traffic is changing, so you can adapt – divert this road, open that gate, and so forth.”

His research aligns with CAIS’s mission to collaborate with domain experts and public policy leaders in seeking equitable technological solutions to priority societal challenges while respecting privacy and data ownership concerns.

“What we're working on now is a universal mobility platform that can integrate all three of these different modalities – hardwired and temporary terrestrial cameras and drones – to give a more complete picture toward mitigating congestion and emissions. If we can make traffic more efficient through better traffic analytics, then we can contribute to the economy by making the transit times of people and goods shorter. The big wins for society are more efficient commuting, lower costs, lower emissions and hopefully better safety, especially for vulnerable road users,” says Elder.

“Working with public sector agencies like the Ontario Ministry of Transportation and innovative Canadian transportation engineering companies, such as Trans-Plan, our goal is to translate this research into real products that improve quality of life for Canadians.”

Making the built urban environment accessible for wheelchair users is something Assistant Professor Mahtot Gebresselassie of York’s Faculty of Environmental & Urban Change is working on. She received a Connected Minds seed grant for her AI and Disability Accessibility in Toronto project as well as a Connected Minds travel grant to work with researchers at Mekelle University in Ethiopia.

She hopes to pilot the project in the Jane and Finch area of Toronto and at York University’s Keele Campus.

“If you're a wheelchair user or a person with other types of disability, the built environment is not made for you, unfortunately.”

Planners, urban designers and architects don't always think about the user, she adds, and when they do, it is not usually a person with a disability. She should know as an architect and urban planner herself.

“Because wheelchairs require space, it’s really difficult for wheelchair users to maneuver the built environment if it is not made with their needs in mind,” says Gebresselassie.

That’s particularly true for pedestrian sidewalks and intersections where things like electric poles, potholes, or a slope that’s too steep, can become barriers. “Wheelchair users should be able to use pedestrian infrastructure like everybody else, but such barriers make it challenging for them.”

Just how accessible sidewalks are for wheelchair users is one of the main questions of her research. “The ultimate goal is to be able to scale this out where it can be used for different neighbourhoods or entire cities.” Using AI provides a quicker, more consistent and less expensive way to audit different areas of the city than potentially hundreds of human auditors doing it manually.

“We used the City of Toronto’s accessibility guidelines, extracting the information for wheelchair accessibility and any other pertinent data to develop an AI model combining it with an aerial map of the Jane and Finch area to see which sidewalks are compliant,” says Gebresselassie, who received a Social Sciences and Humanities Research Council Insight Development Grant for some of her research. The model would also rank streets based on their accessibility.

“Because wheelchairs require space, it’s really difficult for wheelchair users to maneuver the built environment if it is not made with their needs in mind.”

The next step is to develop a smartphone app for wheelchair users that suggests the best routes for a particular destination based on the model’s ranking system. She is doing similar work in the city of Mekelle in Tigray, Ethiopia where she discovered most sidewalks are inaccessible.

Solving mobility challenges and building transportation systems that are safe, inclusive and sustainable are at the core of Professor Gunho Sohn’s research. He is Chair of the Department of Earth and Space Science and Engineering at York University’s Lassonde School of Engineering and the founding director of the Mobility Innovation Centre, which brings together researchers, industry partners and public agencies to shape the future of smart mobility.

As director of the Augmented Urban Space Modelling Lab, Sohn led the creation of a 3D digital twin of York University’s Keele Campus, developed in partnership with Esri Canada and the ISSUM team. As a dynamic virtual environment, it enables researchers to simulate the interactions between pedestrians, cyclists and sidewalk delivery robots in shared spaces. “We know our cities will soon include autonomous systems alongside humans,” says Sohn. “Digital Twin allows us to design for safety, accessibility and community benefit before deployment.”

His work also extends to large-scale, real-world transit systems. As a lead researcher in the Ontario Train Autonomy Collaboration with Thales Canada, he helped develop AI-based perception systems to support safer autonomous rail operations. Sohn also leads the 3D Mobile Mapping AI program – a $2.6-million collaboration with Teledyne Optech – focused on helping autonomous systems understand and navigate their surroundings without relying on GPS.

His team has developed mapping techniques that combine camera and laser sensing to allow vehicles to “see” and move safely through roads, pathways and public spaces. This work provides the spatial awareness that autonomous mobility systems need to operate reliably and safely in real-world environments.

Sohn’s newest project, Smart Mobility Advanced Research & Training (SMART), recently received $1.65 million in funding from the Natural Sciences and Engineering Research Council of Canada’s Collaborative Research and Training Experience program to train the next generation of experts in AI-driven, connected and sustainable mobility systems.

AI, digital infrastructure, mobility policy and community health experts will collaborate with the Opaskwayak Cree Nation (OCN) and the Opaskwayak Health Authority in Manitoba to co-create mobility solutions tailored to community priorities.

“We’re tackling urgent challenges in health, transportation and accessibility – including the smart delivery of fresh food from OCN’s vertical farm to households, supporting wellness and food security,” says Sohn.

The SMART program builds on Sohn’s previous work with digital twin systems and includes real-time simulation and testing, AI-driven traffic optimization, sustainable mobility using electrification, data governance, and autonomous driving and navigation.

As more roads and highways are built or expanded, navigating the chaos whether a person, robot or vehicle, can be complicated. Sohn, Elder and Gebresselassie are working on solutions to ensure people will be moving seamlessly and safely.