����ɫ��

The spaces are small and strewn with rubble. The ground might be unstable and the walls could collapse. It’s hard to tell what else lurks ahead. Somewhere in this collapsed building there are people trapped but it’s not safe yet to send in more people to get them. The clock is ticking.

That’s a scenario trains for. It is one of six teams across Canada that respond to large-scale domestic natural or industrial disasters. Think structural collapses, earthquakes, hurricanes, floods like Calgary’s in 2013 and fires like Fort McMurray’s in 2016. The team includes doctors, paramedics, structural engineers and other first responders. Its urban search and rescue missions could eventually include robots.

“There are two things we are critically short on in any event and it is time and information,” says Simon Bradley, team lead for Canada Task Force 2. Robots can help with both. While Bradley’s team prepares to secure the scene for humans, robots could be sent ahead to gather information on details ranging from air quality to what the collapsed building looks like from the inside. They could also connect with a patient much quicker than the rescue team.

We can put robots in places where we would never put people.

���䲹����������’s at the Schulich School of Engineering is building the type of robots that could be sent into these dangerous scenarios. Dr. Alejandro Ramirez-Serrano and his team are developing search-and-rescue robots that are fast (or, at least, faster than humans), can work in confined spaces, and, in the most challenging circumstances, be fully autonomous.

Train robots to understand the context

“Sometimes, we cannot communicate with the robots when they are deep in the rubble,” says Ramirez-Serrano, PhD. “We want them to make decisions as they see fit, based on the mission and the perceived environment.”

That means training robots to understand context. Much like a firefighter might scan a space and notice if a ceiling is about to cave in or debris could be moved to clear a path, the robot needs to be able to assess the situation, cope with uncertainties and make decisions. Because every situation is different, those decisions can’t always be preprogrammed; the robot needs to think on its own.

That goal hasn’t been achieved yet, but Ramirez-Serrano is working on it. Part of the process is understanding what decisions need to be made and how to achieve them fast in a search-and-rescue situation.

What would an expert do?

“It’s difficult to put the knowledge of an expert into the robot,” says Ramirez-Serrano, a professor in the Department of Mechanical and Manufacturing Engineering. “In an attempt to enable robots to understand the situation, the space, and determine what would the best course of action, we have been interviewing experts and asking them what they would do in a situation. We interviewed many people because … there is typically more than one way to achieve a given mission.

"We don’t want the robot to mimic a specific individual. We want the robot to say, ‘I have all this knowledge, and this is what I have to do.’”

To develop these robots, Ramirez-Serrano’s team works from concept stage all the way through artificial intelligence, perception, control, navigation, user-friendliness and software development. “We work from concept because there are no machines that have been designed to do these things,” Ramirez-Serrano says.

His is the only group in North America and one of only two in the world working on deploying robots in confined chaotic and unstructured spaces for which very little information exists.

Research partnership with local and national agencies

These search-and-rescue robots are an project, a partnership between UCalgary and the City of Calgary to apply research to real-world challenges. The project is partnered with Canada Task Force 2 and the Calgary Emergency Management Agency, and recently received a $500,000 grant (over five years) from the National Sciences and Engineering Research Council. 

For Canada Task Force 2, its main priority in urban search and rescue is improving patient outcomes. If there are people trapped, the task force wants to get them out alive, and the hope is these robots will improve its ability to do that.

“We are not building robots to replace people,” says Ramirez-Serrano. “We are building robots to assist and complement abilities. Search and rescue is dangerous, and humans and robots can work together.”