Winnipeg Free Press: U of M’s humanoid lab recruits new talent
International talent is helping stack the odds for the University of Manitoba’s Autonomous Agents Laboratory in advance of the HuroCup Games.
The team of researchers and programmers in the U of M’s computer sciences department have recruited 21-year-old Qiuting Gong to whip the lab’s humanoid robot, Jennifer, into shape.
Jennifer, the same robot that made local headlines for playing hockey and successfully tackling downhill skiing, will compete at the FIRA Roboworld Cup in Taiwan at the end of the month.
"What we do, in part, is work improving the intelligence in humanoid robotics and one of the ways that we do this is by doing demonstrations and competitions at athletic events," explained John Anderson, head of the computer sciences department.
Jennifer will be competing in about a dozen events, including obstacle runs, penalty kicks, basketball, and archery.
"These are all tasks and techniques that are designed to explore different aspects of intelligence: eye-hand coordination, speed, movement, and planning complex motions," Anderson said.
Gong, an undergraduate student at Harbing Engineering University in Heilongjiang, China, is in Winnipeg working with the lab as part of the federally funded Mitacs Globalink program. Thirty international students are currently in Winnipeg working in different disciplines at the U of M and the University of Winnipeg through Globalink.
During her three month internship, Gong will be refining the programming that allows Jennifer to successfully make a free throw by picking up a ball and throwing it into a basket.
Prior to arriving in Winnipeg, Gong said she had not worked with the computer language that controls Jennifer during her studies in detective guidance at Harbing, but the opportunity to work with humanoid robots was worth learning the new system.
"I needed to learn so much new things," she said. "The last month I was learning and getting myself prepared for the work.
"This is a robot that’s like a man," she added. "It has no life, but I give it life. It’s amazing."
Using a technique called visual servoing — which employs visual sensors to control the robot — Gong is hoping to make Jennifer’s artificial intelligence more dynamic so that the robot can spot a ball from across the court, align itself to the net, and make shots at different distances and angles.
"It’s a challenge for me and that is what I want," Gong said.
Amir Hossein Memar, a PhD candidate at the U of M and Gong’s supervisor in the Autonomous Agents Lab, said the rate of Jennifer’s progress has been increased thanks to Gong’s work.
"Usually we have students who are studying in the university and they come in during the times they are not busy," Memar Hossein said.
"Now that Qiuting is here and since this is her task, she’s been working on it every day and that’s actually a good thing. Because there is more time and she can work and focus more on a given task."
As when people pick up new sports, Hossein Memar said Jennifer’s practice time with Gong is critical to having any success at the upcoming games and the bot’s shooting percentage is increasing daily.
"It’s exactly like a human — if you don’t put in time you won’t get a result," he said.
By: Danielle Da Silva