II-EN: Robotic Equipment for the Investigation of Dexterous Two-Handed Manipulation

Carnegie Mellon University, Pittsburgh PA

Investigators

Abstract

"This award is funded under the American Recovery and Reinvestment Act of 2009 (Public Law 111-5)." One of the grand challenges in robotics is to achieve dexterity, especially dexterity in manipulation. The equipment funded by this project ? a robot torso, two arms, and a custom designed robot hand, supports investigation of manipulation tasks, with a particular focus on two-handed dexterous manipulation. Primary projects include (1) task transfer of two-handed manipulation tasks from human demonstrations and (2) application domain specific designs for a non-dominant robot hand. For task transfer, this project explores a combination of task understanding, planning and learning, and direct user interaction, following along the lines of our previous research in one handed grasping and manipulation. For hand design, this project pursues design optimization to develop the simplest possible robot hand capable of accomplishing a suite of everyday tasks, operating as the non-dominant hand. This approach promises new insights into the required number of degrees-of-freedom of the hand, the required number of drive motors, required sensing, and the geometry of critical but often ignored parts of the hand, such as the palm. Broader impact includes forming a better understanding of human motion. We believe that we can only successfully transfer a task from a human demonstration to a robot performance if we understand a great deal about why a person chooses to perform a task in a certain way. Our proposed robot hand design project also has great implications for prosthetic design, as our robot hand design process will be driven with goals of simplicity in design and ability to perform tasks from everyday life.