Knowledge Bank

Assistive robot hands are used in both medicine and industry. For example, a robot hand can be used as a prosthesis for amputees. Because of the dexterity of a human hand, it has been the gold standard for the imitation of a robot hand. However, because of the complexity of human hands, there is still much we do not know about human hand's control. We still don't know much about whether humans will behave similarly when using assistive device to do the tasks of real human hands. The goal of this research is to understand the kinematics and the control algorithm of humans using robot hands compared to using their own hands. To answer these questions, an assistive two-fingered robot hand with a human interface was designed and fabricated based on the open-source design of the Yale Hand. The control system of the robot hand was created using Arduino. We then performed experiments with human subjects to collect kinematic data of the subject's hand, arm, and robot hand under two conditions: (1) when subjects use the robot hand to pick and drop objects, controlling the robot hand using a joystick, and (2) when subjects use their own hands to pick and drop objects. We performed dimensionality reduction via Principal Components Analysis to characterize the relative complexity of the human actions with their own hands versus with the robot hand. Our results showed that the number of principal components of hand-based picking/dropping was lower than that of robot-based picking/dropping. So, the kinematics of hand-based picking/dropping was less complex than that of robot-based picking/dropping. We suggest that this is because while using the robot hand, human needed to compensate for the low degree-of-freedom of the robot hand and their lack of expertise in using the robot hand. This research project will advance the knowledge of humans using assistive robots and improve human living by applying them in designing more robust assistive robot hands.