If you find yourself near Maple Creek, Saskatchewan, you may notice a large structure above the train tracks. This is a machine vision inspection system, called the Train Inspection Portal System (TIPS). The system is designed to enhance the safety of railcars on freight trains by detecting defects such as cracks in bearings and wheels, or wear and tear on couplers and draft gears.
As trains pass through this portal, more than 35 infrared cameras capture 72 high-resolution images of each railcar from every angle. These images are sent to remote offices where trained employees inspect them and relay any issues to field workers, who ensure that repairs are made promptly when the train reaches a station or yard.
Traditionally, certified inspectors walked around stationary trains to spot defects. However, machine vision technology improves this process by identifying safety issues while trains are in motion, including hard-to-see areas such as the undercarriage.
Canadian Pacific Kansas City (CPKC) railway is a leader in the use of vision technology for railcar inspection. Their TIPS was chosen by Rail Research Advisory Board of Canada (RRAB), a collaborative rail R&D forum of industry, government and other stakeholders, for a pilot study to evaluate the benefits and challenges of this technology. The project, funded by Transport Canada, allowed us to work with RRAB on adapting railway regulations to incorporate safety improvements through technology and innovation.
"We wanted to validate a better process for improving safety that could ultimately help railroads expand the services they provide to customers," says Kyle Mulligan, Assistant Vice-President, Operations Technology Team, CPKC.
As part of the study, Transport Canada engaged the National Research Council of Canada (NRC) to lead the performance evaluation of TIPS's hardware and software. "The NRC is an unbiased third-party that can conduct the research independently but in collaboration with other stakeholders," says Kaveh Khalilian, research development officer, Transport Canada Innovation Centre. "They also bring in specialized expertise to customize the research to our needs and those of the railway." The NRC's expertise on this project included our specialized skills in railway engineering, data analytics and machine vision technologies.
Alireza Roghani, research officer at the NRC's Automotive and Surface Transportation Research Centre, says this became a collaborative effort between the NRC, CPKC, Transport Canada and the University of Alberta. "We developed a comprehensive scope for the research that included field experiments, historical data analysis and extensive image scrutiny," he says.
The NRC involved the University of Alberta to conduct human factor analysis and field experiments. "Car inspectors sitting in the control centre needed to be taught how to interpret thousands of images as train cars roll through the portal at track speeds up to 100 kph," says Lianne Lefsrud, Risk, Innovation and Sustainability Chair and professor at the University of Alberta. "We had to identify factors that might lead to human error or affect their ability to do the job well."
The study found that the most significant advantage of using vision technologies is the ability to detect issues in the undercarriage, which are often difficult for human inspectors to see. However, challenges such as image quality being affected by snowfall and other environmental factors were also identified.
Further improvements to the system could involve incorporating AI algorithms to reduce undetected defects and enhance the capabilities of remote safety inspectors. AI will also help railways manage a more diverse range of fleets, cars and cargo.
"This collaborative project with CPKC, the NRC and the University of Alberta is an excellent example of how large organizations can work together to evaluate technology that can be integrated into future operations industry wide," says Khalilian. "The technology improves innovation, efficiency, productivity—and the reliability and resiliency of our transportation system as a whole." All factors that can positively impact the Canadian economy.