Inspecting bridges without the risk
Bridges in Canada require regular inspection to keep the public safe; however, it can be an arduous and sometimes dangerous task. Professor Alex Ferworn is working on an unmanned solution. (Photo by Valor Kopeny)
Building on his experience using robots and drones to examine search and rescue sites, computer science professor Alex Ferworn is applying the technology to facilitate one of the tasks that helps keep Canadians safe on the roads: bridge inspection.
Professor Ferworn’s past research has included successful applications of drones in machine learning. His Disaster Scene Reconstruction (DSR) project uses a drone to carry a RGB-D camera to capture the true, 3D form of building rubble in a point cloud. The data is then run through a game engine in order for first responders to make decisions about how to best carry out tasks at a disaster scene.
His most recent project will use drones to assess the condition of bridges for structure and integrity. Bridge inspections are typically conducted on site by a structural engineer, but it carries the potential risk of falling or injury. Using a DGI Phantom Drone equipped with a 3D camera, professor Ferworn’s team is able to safely and remotely inspect bridges using digital technology. “We are looking for breaks, excessive corrosion and other damage,” he said.
Professor Ferworn is collaborating with a cross-disciplinary team of students, including a PhD student in aerospace engineering and master’s students in computer science. Together, they will incorporate new capabilities in future unmanned aerial systems (UAS) designs, but their work may also have impact on how sensing, imaging and wireless transmission technologies can be combined in automatic services.
The data collected will also form a framework around which the bridges can be assessed over time as historical data can be compared to fresh data. “Through this work, the process of routine inspection could be automated and be done more often, allowing the safety of bridges and similar structures to be ascertained more reliably,” said professor Ferworn.
At this stage in his research, the drones are being flown manually. However, over time, professor Ferworn anticipates being able to make the process fully automated using footage already recorded by the drones and applying post-processed machine vision techniques to map out paths.
Additionally, two students under professor Ferworn’s supervision are undertaking their own research into the use of drones. Dalia Hanna’s project uses UAS to predict the paths of lost and wandering dementia patients through algorithms generated from the knowledge collected by the drones. Christopher Chan’s project simulates UAS-carried improvised explosive devices (IEDs) in order to better detect threats and thwart potential attacks.
Professor Ferworn’s work in this field began as a way to perform search and rescue safely and quickly by augmenting efforts with the assistance of robots and imaging equipment. Since then, his work has evolved, using robots and unmanned aerial devices to tackle a variety of tasks that minimize risk of human injury while increasing speed, efficiency and accuracy.
This project is supported by a grant from the Natural Sciences and Engineering Research Council of Canada.