This award is funded under the American Recovery and Reinvestment Act of 2009 (Public Law 111-5).
This project is defining the basis for lower-complexity robotic hands that can grasp a wide variety of objects in noisy and unstructured environments. The new generation of mobile and humanoid robots still lacks basic ?hands? that can reliably grasp objects. Robot hands have been traditionally built as anthropomorphic, high degree-of-freedom (DOF) mechanisms that are expensive and difficult to control. The research team is developing technologies based on defining hand mechanisms that capture two key features of human grasping, versatility and low dimensionality of hand postures. Reducing complexity brings major benefits. Determining the minimal number of hand joints, sensors and actuators can reduce costs and speed research as low-complexity hands can be easily fabricated, designs can be quickly iterated, and control can be simplified. These ideas are used to build a low-cost, low DOF grasping device that is based on hard human grasping data. Further, the new hand designs are being tested in simulation so as to build hardware that is functionally proven for robotic grasping tasks. Important research outcomes include: development of a new low-dimensional, low-cost robotic hand; experiments to gain insights from human grasping and adaptive compliance; and machine learning algorithms for grasping. Broader impacts include: collaboration between neuroscience and robotics; hardware design methods and computational tools for hand researchers; providing robust grasping capabilities in real environments such as robots for home care and assistance for the elderly and disabled; establishing links between neural control and prosthetic devices based on dimensionality reduction; and dissemination of modeling and simulation grasping software.
