IEEE Robotics & Automation Magazine - March 2013 - 81

Table 5. User-acceptance questionnaire.
Statement

Construct

Minimum Maximum Mean

Standard
Deviation

1) I think it's a good idea to use the robot.

Attitude toward using the robot
( a = 0.64)

1.0

5.0

4.53

0.89

2) The robot would make my life more interesting.
3) I would use the robot again.

Intent to use

2.0

5.0

4.53

0.94

4) I think the robot can help me with what I need.

Perceived adaptability

1.0

5.0

3.59

1.41

5) I enjoy the robot talking to me.

Perceived enjoyment

4.0

5.0

4.65

0.49

6) I find the robot easy to use.

Perceived ease of use

2.0

5.0

4.53

0.79

7) I find the robot pleasant to interact with.

Perceived sociability
( a = 0.50)

1.0

5.0

4.37

0.96

Perceived usefulness
( a = 0.84)

1.0

5.0

3.44

1.50

Social presence
( a = 0.62)

1.0

5.0

3.46

1.39

Trust
( a = 0.86)

1.0

5.0

3.53

1.32

8) I feel the robot understands me.
9) I think the robot is nice.
10) I think the robot is useful to me.
11) It would be convenient for me to have the robot.
12) I think the robot can help with many things.
13) When interacting with the robot, I felt like I'm
talking to a real person.
14) It sometimes felt as if the robot was really looking
at me.
15) I can imagine the robot to be a living creature.
16) Sometimes the robot seems to have real feelings.
17) I would trust the robot if it gave me advice.
18) I would follow the advice the robot gives me.

A number of participants also engaged in open dialogue
about the robot with our research team.
Results and Discussion
During the two days, we were able to observe the interactions of 40 elderly participants having mild Alzheimer's
disease, mild cognitive impairments, and normal cognitive control. We also obtained 22 completed questionnaires from participants ranging in age from 57 to 100
years old. For this exploratory study, we did not categorize
the results for different cognition levels because of the
short-term nature of the interaction but rather focused on
how the older adults interacted with a humanlike robot to
obtain feedback to optimize the robot design.
User engagement was observed for each interaction of an
activity to identify the presence/nonpresence of the previously mentioned engagement indicators (manipulation of
objects and attention to the activity/robot). An interaction
was defined as including the robot detecting the user's action
(which updates the activity state) as well as the corresponding reaction from the robot. The results were then categorized into a participant being engaged: all the time (constant
presence of engagement indicators), some of the time (at
least one or more instances where nonpresence of engagement indicators is detected), or none of the time (constant
nonpresence of engagement indicators). A similar approach
was used to determine compliance. The results showed that
33 participants were engaged all the time and seven were
engaged some of the time. With respect to compliance, 35
participants complied with the robot all the time, four par-

ticipants complied some of the time, and one participant did
not comply. Hence, the majority of participants both
engaged and complied with Brian 2.1. The participant who
did not comply with the
robot stated that its voice
To promote the social
was causing interference
with his hearing aid. Since
dimensions of an activity,
the robot was placed in a
large space with a lot of
Brian 2.1 greets a person,
background noise, we
used an amplifier to
tells jokes, and provides
increase the volume of the
robot's voice. Although
general positive statements
the participant had this
issue, he sat with and
about the interaction
watched the robot for
approximately 5 min
or the activity.
before letting one of the
research team members
know about the interference. The average length of time participants interacted with
Brian 2.1 was 12.6 min. There were no observable differences between the two activities for engagement and compliance during the interactions.
We also observed participant reactions to both the
happy and sad emotional behaviors of Brian 2.1. When
the robot was happy, we found that 82% of the participants either smiled back at the robot or laughed. For the
seven participants who were distracted at least once during the interactions, at which time the robot displayed a
March 2013

IEEE ROBOTICS & AUTOMATION MAGAZINE

Table of Contents for the Digital Edition of IEEE Robotics & Automation Magazine - March 2013