Sensory Robotics blurs lines with human capability

Thursday, September 21, 2017

Assorted hands and arms are starting to overtake Soo Jeon’s office – but he doesn’t mind in the least.

A professor of mechanical and mechatronics engineering at the University of Waterloo and RoboHub research affiliate, he is focused on finding a solution to a key issue in the burgeoning field of robotics.

“In terms of their hardware, progress has been good,” Jeon says. “The physical capabilities meet and exceed human capability. But robotics aren’t as advanced in perception and dexterity. The question is: how do we use sensory data to design controls in a smart way?”

Groundbreaking Research in Sensory Robotics technology

As he continues the search for an answer, people outside the University are taking notice of his ground-breaking research.

In 2015, the Natural Sciences and Engineering Research Council of Canada (NSERC) awarded Jeon a $120,000 Discovery Accelerator Grant to support his work on sensory integration and motion control.

Testing out the equipment are Soo Jeon, and his graduate students, middle Nan Wei, a masters student and PhD student Shiyi Yang, seated.

Testing out the equipment are Soo Jeon, and his graduate students, middle Nan Wei, a masters student and PhD student Shiyi Yang, seated.

NSERC selected Jeon for a Research Tools and Instruments Grant, which allowed him to purchase two Proficio robotic arms used in rehabilitation therapy for stroke survivors. Working with the artificial limbs, Jeon and his team discover how to design robots that can sense and manipulate objects as humans do.

“We are talking about designing robots that can attach to the wheelchairs of people who are paralyzed, so they can pour and drink a glass of water by themselves, for example,” Jeon says. “We need robots that can work inside homes and hospitals, and robots that can be used commercially to work alongside people, doing the more strenuous or dangerous jobs.”

According to Jeon, these applications require robots that are both agile and adaptive.

“Conventional robots work in structured factory settings, but it’s very dangerous for people to be close to those robots,” he says. “To work side-by-side with people, these robots will have to be able to sense a person approaching or an object showing up in the workspace, and then adapt accordingly. Designing these control techniques is the challenge.”

Reverse engineering for advanced robotics

One of the ways Jeon is tackling these challenges is by working with researchers outside of the mechatronics engineering department.

A colleague, James Tung, introduced him to Waterloo kinesiology professors Michael Barnett-CowanEric Roy and Ewa Niechwiej-Szwedo, and they are helping him understand why humans are so dexterous compared to robots.

“As babies, we learn how to control our hands and manipulate objects, and then we store that learning in our memory,” Jeon says. “From then on we are able to more or less instantly estimate the properties of objects around us and manipulate them. This process isn’t easy to replicate in robotics.”

Although the work ahead of him will be challenging, Jeon says he is confident he’s in the right place to do it.

“The University of Waterloo values and promotes innovation,” he says. “Being here allows me to focus on my work, teach my students and make progress on my research.”

This article was originally featured on the Engineer of the Future site.