Page 36 - Human Inspired Dexterity in Robotic Manipulation

32 Human Inspired Dexterity in Robotic Manipulation

made of white plastic and a vertical handle mounted on the end of the base.
The handle was made of gray plastic and equipped with two hidden six-axis
force-torque sensors. The design of the handle enabled measurement of
forces and torques applied by the digits. The design of the object provided
visual geometrical cues that were congruent with the objects’ mechanical
properties, such as mass and mass distribution [31]. Unlike previous studies
using objects that gave conflicting visual and physical information, the
congruent visual geometrical cue here was used to allow subjects to be aware
of the task context using object orientation (i.e., handle on right or left).
Motion capture system was used to measure object peak roll and lift onset.
Technical details of the apparatuses and force/torque data processing
algorithms can be found elsewhere [29,31,32].

3.2.3 Protocols
Subjects sat in front of the object with the arm positioned in a way that
ensured comfortable grasp and lift actions. On hearing a “go” signal, subjects
reached to grasp and lifted the object above the table, held it in a stationary
position for 2 s, and replaced it on the table. Subjects were required to
grasp the specified handle with the tip of the thumb on the left contact sur-
face and the tip of the index and middle fingers on the right contact surfaces
of the handle, and to prevent the object from tilting. Subjects then replaced
the object back on the table. All tasks required subjects to plan and generate
torques in an anticipatory fashion to compensate the torque caused by the
asymmetrical mass distribution of the object. Additionally, subjects were
instructed to rotate the object 180 degree after several lifts (Fig. 3.1A).
Therefore, the two task contexts differed based on opposite cues induced
by the object orientation, which required opposite compensatory torque
(T com ) production from the hand. Every time the object was rotated, the
context switched.
Before the experiments started, the object was visually presented to the
subjects and they were allowed to briefly touch the graspable surfaces to
familiarize themselves with the frictional properties of the handle. All exper-
imental conditions consisted of four blocks of eight consecutive trials with
the exception of the Random condition (see later on) and each block was
performed in the same context. As there were two manipulation contexts
(i.e., left or right center of mass), the order of presentation of R and
L contexts was counter-balanced across subjects for each experimental