Bretl, McCarthy win Best Manipulation Paper at IEEE Robotics and Automation conference

2/21/2013 Kim Gudeman, CSL Communications

CSL Assistant Professor Timothy Bretl and former ECE undergraduate student Zoe McCarthy have won the Best Manipulation Paper Award at the IEEE International Conference on Robotics and Automation. Their paper was chosen out of a record 2,032 submissions.

Written by Kim Gudeman, CSL Communications

CSL Assistant Professor Timothy Bretl and former ECE undergraduate student Zoe McCarthy have won the Best Manipulation Paper Award at the IEEE International Conference on Robotics and Automation. Their paper was chosen out of a record 2,032 submissions.

Assistant Professor Tim Bretl received a 2010 NSF Faculty Early Career Development (CAREER) Award worth $400,000. He will use the award to create novel prosthetic devices.
Assistant Professor Tim Bretl received a 2010 NSF Faculty Early Career Development (CAREER) Award worth $400,000. He will use the award to create novel prosthetic devices.
Timothy Bretl

The paper, “Mechanics and Manipulation of Planar Elastic Kinematic Chains,” provides a mathematical model for solving a problem that has mystified researchers for years: How to enable robots to manipulate deformable, or flexible, objects.

While it is comparatively easy to program a robot to manipulate a rigid object, such as a pencil, researchers have found it much more difficult to instruct a robot to pick up a wire or string. That is because it is more challenging to mathematically describe a deformable object’s position, orientation and shape, as they are almost infinitely changeable.

“For people, these problems aren’t hard. We play with paper like we play with pencils,” said Bretl, an assistant professor of aerospace engineering. “But it seems very hard to give a robot a concise description of all the possible shapes that a deformable object can take.”

Bretl’s team discovered that by modeling the shape of a deformable object as the solution to an optimal control problem, and by studying the geometry of this problem, it became simple to describe all possible shapes. This result led to an algorithm for manipulation planning that was easy to implement and that performed well in practice.

In the future, the work could be applied to improve robotic surgery, which critically depends on the manipulation of deformable objects like surgical thread and tissue. In addition, robots using this methodology could be programmed to place car mats in automobile manufacturing plants – one of the few processes that has yet to be automated, due to the complexity of picking up, folding and placing deformable mats in the cars.

“The beauty of this work is that while the analysis was really hard, the results are actually really easy to apply,” Bretl said. “There’s definitely more that needs to be done, but we have provided a solid foundation for future efforts.”


Share this story

This story was published February 21, 2013.