Page 60 - Human Inspired Dexterity in Robotic Manipulation

56 Human Inspired Dexterity in Robotic Manipulation

The slave is a dummy hand. The MCP joint of the index finger is actu-
ated by a DC servo motor. The hand is covered by a rubber sheet to hide the
internal components. Soft urethane foam is stuffed inside the index finger
and the thumb allowing the fingers to behave like human soft tissue and
to be seen as natural from the subject. The slave is roughly the same size
as the average human hand and is approximately 280 g in mass.
The system was implemented as a unilateral one degree-of-freedom
MSS, by having the measured master MCP joint angle fed back to the slave
MCP joint. The system was built using a real-time control OS (Matlab xPC
target, MathWorks, Inc., USA), thus the time delay between the master and
slave can be artificially controlled. The system was designed to perform a
simple table-tapping task using the index finger. Because the task involves
self-motion, it is an extension of the classic RHI. Haptic sensation is pro-
vided by the device through the user’s exposed index finger touching the
table surface. However, the master is not equipped to deliver force feedback.
To investigate multisensory illusions with the developed MSS, we con-
ducted two experiments. In the first experiment, three trials (a, b, and c)
were conducted to compare multisensory illusions elicited in different con-
ditions. Trial a involved a classic RHI, Trial b included table tapping, and
Trial c involved simultaneous table tapping and brush stroking. All trials
were performed by 14 subjects, all of whom wore the master on the right
hand. Each trial included 5 min of stimulation for the subject. To assess
the multisensory illusion, we used proprioceptive drift and questionnaire
assessments. Proprioceptive drift is a psychophysical measure that can be
measured by a series of intermanual reaches with eyes closed. Before and
after the stimulation, the subject is asked to point using their left index finger
at the position of their right index finger extended to the opposite side of the
table. The indicated position is known to be displaced leftward toward the
dummy hand, and the magnitude of drift distance is proportional to the
duration of the illusion [10]. The questionnaire was configured based on
the previous study by Botvinick [10] that consists of five statements
(Q1–Q5), including the predicted phenomena (Q1–Q3) and the unpre-
dicted phenomena (Q4 and Q5). Each subject was asked to answer the ques-
tionnaire with a score from 0 (lowest) to 5 (highest). The reference [12]
describes the experimental procedures in detail.
Fig. 4.3 shows the experimental results of proprioceptive drift in each
trial. Larger proprioceptive drift suggests stronger multisensory illusion
occurring. It was revealed that self-motion significantly enhanced the illu-
sion (P < .001*** two-tailed paired samples t-test). The questionnaire
assessments supported this result.