GRAB Lab, Yale University

While working in Aaron Dollar’s GRAB Lab at Yale University, I led the development of a dexterous 6-degree-of-freedom robotic hand based on the common Stewart-Gough Platform parallel mechanism. The hand consist of six prismatic actuators used for manipulation and one revolute actuator (in the palm) used to close the fingers. While traditional robotic hands require significant sensing and feedback to manipulate objects, the simple inverse kinematics of the parallel mechanism underlying this hand allow it to precisely manipulate objects without any closed-loop feedback or prior object geometry knowledge. 

In order to optimize the performance of the hand, the manipulation workspace was numerically predicted considering kinematic and frictional constraints. Dimensional synthesis was then performed to optimize the hand’s geometry to yield the largest range of manipulation. These results were then validated experimentally.