Page 109 - Human Inspired Dexterity in Robotic Manipulation

Approaching Human Hand Dexterity Through Highly Biomimetic Design 107

Fig. 6.16 Snapshots showing the teleoperation process of our proposed
anthropomorphic robotic hand. Top row: The index and middle fingers can move
independently from the coupled ring and little fingers. Middle row: The precision
grasp between the index/middle finger and the thumb. Bottom row: The abduction,
adduction, and opposition motions of the thumb.

information about how much the tendons should travel back and forth, the
control problem of our proposed robotic hand can be greatly simplified by
bypassing the complicated mapping conversion process that often involves
the nonlinear relationships in the m n Jacobian matrix.
As shown in Fig. 6.15, our data glove uses six custom-made string poten-
tiometers whose original design was adopted from the online open-source
STL files [18]. Each string potentiometer works as a key retractor with an
anchor point at the corresponding tendon insertion site near the distal joint
of each finger. It can measure the length of the string when it travels back and
forth along with the finger movement.
As demonstrated in Fig. 6.16, the data glove was successfully used to
teleoperate our anthropomorphic robotic hand via an off-the-shelf servo
controller (CM-530) [17]. During this process, information of natural hand
motions from the human operator were extracted by the data glove and
transformed into data about how much each string (tendon) should travel.
These data were simultaneously converted to rotational angles of each servo
based on the diameter of the pulley directly connected to its shaft.

6.5 PERFORMANCE OF THE BIOMIMETIC ROBOTIC HAND

To evaluate the efficacy of our proof-of-concept design, in this section, we
first quantitatively investigate the fingertip trajectories, and then