AI in trucking project aims to build national R&D data vault

Researchers at the University of Manitoba and the NRC are using AI to model trucking routes across Manitoba, which they say could reduce the risks of disruption from warming winters, extreme weather events and flooding.

Dr. Jon Regehr, an associate professor of civil engineering at the University of Manitoba and one of the researchers for the project, says the work will allow governments to better make decisions about where to invest in infrastructure and build resiliency in the system.

"The early days of the pandemic actually highlighted this question a lot," Regehr said in an interview with Research Money. "We wrote the proposal before the pandemic, but the issue is the same. Suddenly, a disruption occurs, a border closes, and nobody has toilet paper or whatever product — it just doesn't show up."

The six-year project is a collaboration between the university, the National Research Council (NRC), Saskatoon-based company International Road Dynamics and Manitoba Infrastructure. By the end, they will build a mobile trailer to gather data, model potential risks to the province's trucking routes, and develop a data vault for future research on freight trucking.

Regehr and Dr. Babak Mehran, an associate professor of civil engineering at the University of Manitoba, are collaborating with Dr. Chaouki Regoui at the NRC's Digital Technologies Research Centre to collect the data and model risks to the province's trucking network.

The project has received $1.6 million through the NRC's AI for Logistics Supercluster Support program, according to the NRC's website.

The project will produce a new national data vault for future R&D

The first part of the project will be to build a mobile trailer that the researchers can use to collect data, including temperature, weather conditions, the number of trucks using the roads, the types of vehicles on the road and other relevant information. Regehr said it will fill in gaps where there is no data for parts of Manitoba's road network and allow the team to gather precise information about the conditions at different locations.

Once the data is collected, the team can then begin to model potential scenarios, see where disruptions affect traffic the most, and identify where improvements may be needed.

"A fair example is a road prone to seasonal flooding," Regehr said. "How much traffic is it currently handling? How important is that network?"

By using AI techniques to analyze the data, they can look across risks — such as flooding, severe weather or closing due to accidents — and then see which roads are the most vulnerable and most in need of improvement, according to Regehr.

The project is laying the groundwork for further research. For example, the project won't be looking into future climate scenarios or wildfires, but they could be the subject of further research using the data vault that they develop, Regehr said.

International Road Dynamics, the private sector partner for the project, will be providing its expertise in sensors and AI modelling. Rish Malhotra, the company's president and CEO, said that the work being done now lays the groundwork to support electric vehicles and autonomous vehicles as well as more efficient toll roads.

"This is really the starting point to provide information and data from the infrastructure to the vehicle," he said in an interview.

Malhotra added that the data vault will provide three benefits: it allows better decision-making on road networks by governments in Canada, helps to identify opportunities to invest in infrastructure and provides the ability to use AI to model road networks.

NRC program focuses on AI to strengthen logistics systems

The program has separate funding and projects from the Scale AI supercluster in Montreal, but supports the goals and priorities of the supercluster. Like the superclusters, the program facilitates collaborations between universities and the public and private sectors to advance research.

It is one of five supercluster support programs at the NRC, each of which support one of the larger superclusters. The other four programs are:

• Advanced Manufacturing (supports the Canada's Next Generation Manufacturing Supercluster)
• Digital Health and Geospatial Analytics (supports Canada's Digital Technology Supercluster)
• Ocean (supports Canada's Ocean Supercluster)
• Sustainable Protein Production (supports Canada's Protein Industries Supercluster)

The NRC support programs have "no financial overlap" with the superclusters but they have a "shared common vision," said Margaret McKay, the program leader of the AI for Logistics Supercluster support program, in an interview.

McKay said that the NRC, and the Canadian government more broadly, has an interest in building more resilient road networks.

"Having empty containers in the right place is a challenge. And if your empty container is stored in the wrong place, you have to move it to the right place. There is just a lot of things to keep track of, all of which seem very easy [on an individual basis]... but when you look at them at a system level and you want to optimize a system as big and complex as Canada's transportation network, AI is your best chance at that," she said.

The national data vault will be a "playground of different data types" for researchers interested in solving logistics problems or companies looking to develop a product, she added. The idea will be to build a set of historical data about road networks that can be used to make predictions — much like humans do.

"We all drive with a historical model," she said. "I don't try to cross Ottawa at 8:30 in the morning, right? This just does it on a much grander and much more refined scale."

R$