A three-year research project led by Saskatchewan Polytechnic’s Digital Integration Centre of Excellence (DICE) developed an AI-based tool to identify high-probability locations of gravel deposits across the province, potentially saving millions in infrastructure costs.
In partnership with geological experts at Peter Lucas Project Management Inc., the applied research project used machine learning and neural network algorithms to analyze data, including satellite imagery, vegetation data, water tables, lithology and other open-source datasets. The result was a predictive mapping tool that identifies high-probability locations of gravel deposits across Saskatchewan, with testing already underway.
“We are currently field testing it in our partner company, McIntosh Sand & Gravel,” said Connor Regier, CFO of Peter Lucas Project Management. “What we can say is that it’s meant to cut exploration time way down. Work that used to take weeks can now be done in days. That’s the real payoff, and we’re just getting started putting it to use.”
Saving time, money and miles
Gravel is a non-renewable resource essential to construction and road maintenance. Given Saskatchewan’s vast road network – more than 165,000 kilometres of grid roads and over 190,000 kilometres of rural roads – the province relies on gravel to maintain its infrastructure. The mapping tool enables faster, more informed decisions when bidding on large-scale construction projects and can help reduce costs by minimizing travel required for a project.
“A large portion of the cost of aggregates is the hauling costs, so the closer we can find a usable pit to a project, the less trucking is needed – and that’s usually the [highest] cost in supplying aggregates. Less hauling means lower costs, less impact on the environment and quicker turnaround on bids,” said Regier. “Saskatchewan has a huge network of roads to maintain, so even small savings per project add up in a big way.”
Traditionally, gravel exploration relied on manual surveys, aerial photo analysis and electromagnetic testing, a time-consuming and labour-intensive process. The new system replaces much of that work, reducing exploration time from weeks to days. “It’s been a mutually beneficial relationship,” said Regier. “DICE brings the machine learning and AI expertise, and we bring the gravel expertise. Together we have created a great solution.” The technology itself could also be adapted to other resource exploration activities. While the data and AI models would need to be tailored, other potential examples include identifying deposits of specialized sand used in glass manufacturing or predicting where valuable minerals, such as gold, may be deposited.
“Work that used to take weeks can now be done in days.”
– Connor Regier, Peter Lucas Project Management
“The tool works by learning what gravel deposits tend to look like from the outside – the shape of the land, the vegetation – and then spotting those same patterns elsewhere. That same idea could work for other surface-based resources, too,” said Regier. “That said, this version was built and trained just for gravel, so using it for something else would mean starting over with new information, not just pointing today’s tool somewhere new.”
The collaboration began with a small exploratory grant from the Natural Sciences and Engineering Research Council of Canada (NSERC). The final phase of the project received $300,000 in NSERC funding, supporting both the initial research and the development of the final product.
“This is a powerful tool for Peter Lucas,” said Terry Peckham, director and research chair for DICE. “It’s not just about finding gravel. It’s about transforming how we manage a critical resource in a province that depends on it.” Over time, the tool improves as new data is collected and incorporated into the model. “This project took a few years of research, and the tool kept getting better the more we tested and refined it,” said Regier. “Going forward, the plan is to keep feeding it more information to make it sharper still. So, it’s meant to keep improving, not stay the same forever.”
While DICE worked on this collaboration with Peter Lucas, it has also been active in finding other solutions that will ultimately benefit the heavy construction industry. “DICE is working on a critical mineral exploration project with EarthEx and Red River College Polytechnic to advance drone swarming operational capabilities for geophysical surveys,” said Peckham. “The partners are working on co-ordinating the operation of multiple drones under a single operator, reducing the need for large field crews in remote and northern environments.”
Two Saskatchewan Polytechnic alumni played key roles in creating the predictive mapping tool. Deshan Heenatigala, a Computer Systems Technology graduate, worked as a research technologist with DICE and Tai Nguyen, a graduate of the Cloud Computing and Blockchain and Artificial Intelligence and Data Analytics post-graduate certificates, contributed as a student assistant.
DICE provides innovative solutions for solving data and digital challenges for business, industry and non-profits.
Martin Charlton Communications is the largest public relations company in Saskatchewan and one of the largest in western Canada.