Troy, N.Y. - In a project sponsored by the National Science Foundation (NSF), researchers at Rensselaer Polytechnic Institute and the University of Pennsylvania have developed a new framework to support the design of automated manufacturing and robotic systems.

The researchers recently won the Kayamori Best Automation Paper Award when they presented their work at the International Conference on Robotics and Automation, sponsored by the Institute of Electrical and Electronics Engineers.

"Previously, designers of automated manufacturing and robotic systems developed prototypes using intuition gained by trial and error. With the new framework, designers have a precise set of guides to use in prototyping and refining automated manufacturing systems and robotic devices," said Jeff Trinkle, professor and chairman of the Rensselaer Department of Computer Science and one of the authors of the paper.

The new level of precision in the design process potentially reduces the time and cost of designing automated manufacturing and robotic systems used for many purposes, including assembly, part-feeding, and material handling, Trinkle said.

Trinkle worked to develop the framework with Jong-Shi Pang, the Margaret A. Darrin Distinguished Professor in Applied Mathematics at Rensselaer; Vijay Kumar, applied sciences professor at the University of Pennsylvania School of Engineering; and Peng Song, postdoctoral research fellow at the University of Pennsylvania.

"Improvements in the design process for automated and robotic systems are particularly important in facilitating the production of smaller components and devices that cannot be easily or efficiently assembled," said Rensselaer Dean of Science Joe Flaherty. "Solving real-world problems is the goal of Rensselaer's programs in science and computer science."

The framework also can be used to guide the design and control of robots to do a variety of jobs that humans cannot do or that pose undue risks, Trinkle said, such as entering a collapsed building to photograph the damage and determine how to rescue people trapped within; or undertaking missions for the National Aeronautics and Space Administration to repair and service space stations and telescopes and explore planets.

"To plan a task, one must first be able to reliably predict the results of actions," Trinkle said. "For example, if a robot pushes against a box on the floor, we should be able to predict if the box will stick, slide, or tumble. This framework allows us to make such predictions so that we can now think about designing a series of actions to accomplish a specific goal such as,'Put new staples in the stapler,'" he said.

"The new framework uses differential equations and inequalities to predict the motions involved in automated and robotic systems," Trinkle said.

In the future, Trinkle and his colleagues plan to expand the framework, developing theory and computational tools, and test it on more applications. They will work with undergraduate, graduate, and postdoctoral students at Rensselaer and the University of Pennsylvania, and researchers at Sandia National Laboratories in Albuquerque, N.M.

The paper describing the new framework is available at http://www.cs.rpi.edu/~trink/Papers/STKPicra.pdf.

Contact: Robert Pini
Phone: (518) 276-6050
E-mail: pinir@rpi.edu