Using the Negative Attitude Towards Robots Scale with by ghkgkyyt


									         Using the “Negative Attitude Towards Robots Scale”
                     with Telepresence Robots
             Munjal Desai, Katherine                                 Henriette Cramer                      Nicander Kemper
              Tsui, and Holly Yanco                             SICS & Mobile Life Centre                University of Amsterdam
                       UMass Lowell                     
                       {mdesai, ktsui,

1.    INTRODUCTION                                                                  on the shoulder and gave her a “high five.” In the condition
   As robots become more commonplace in society, it is im-                          without the physical contact, the woman and Robosapien
portant to understand how people feel about these robots,                           only conversed. The style of the Robosapien helping the
including how they look, how they behave, and their pur-                            woman was also manipulated; the robot either helped when
pose. In 2004, Nomura et al. developed a psychological tool                         asked (reactive) or offered advice (proactive).
to measure people’s anxiety towards robots which evolved                               The participants then answered questions relating to how
into the “Negative Attitude Towards Robots Scale” (NARS)                            familiar they were with the robot in the video, to what ex-
[6]; the researchers have conducted several user studies re-                        tent they thought the person in the video should follow the
fining NARS and its subscales (e.g., [1, 5, 6]). NARS has                            robot’s advice, and to what extent they would follow the
three subscales: S1 measures “situations of interaction with                        robot’s advice if they were the person in the video. The
robots,” S2 measures “social influence of robots,” and S3                            participants also were asked in what relationship they would
measures “emotions in interaction with robots” [7].                                 view the robot to have in the video if they were the person
   NARS’s frequency of use within the human-robot interac-                          in the video (e.g., friend, assistant, co-worker, pet, acquain-
tion community has been increasing; a sampling of studies                           tance, toy, family).
is shown in Table 1. In the full paper, we will include the re-                        At the end of the survey, participants were asked a subset
maining studies omitted for length and describe each of these                       of the NARS S1 subscale questions with the original word-
studies in more detail. In all of the studies, the participants                     ing from [7] and modified S3 subscale questions in which all
perceived the robots to be autonomous. In this paper, we                            of the questions were altered from their original wording.
examine if NARS can also be used for robots which clearly                           The participants rated each statement on a modified Likert
have a human involved in the operation of the robot. For                            scale from 1 (strongly disagree) to 7 (strongly agree) with
example, telepresence robots can be thought of as mobile                            4 as “neither agree nor disagree.” For analysis of the survey,
embodied video conferencing systems with live video and                             participants were divided into those having a positive atti-
                                                                                    tude towards robots (NARS score below X=3.4, SD=1.0)
audio communication. In a sense, the person operating the
                                                                                    and those with negative attitudes (NARS score above X).    ¯
telepresence robot is using it as a physical robot avatar. In
the state of the art telepresence robots, an operator must                             We examined the internal consistency of the NARS ques-
log into a robot and manually drive the robot around.                               tions’ responses by computing the Chronbach alpha value [4]
                                                                                    for each of the subscales. Given the participants’ responses
                                                                                    to this survey, we computed the Cronbach alpha to be 0.85
2.    NARS IN A ONLINE VIDEO SURVEY                                                 for the S1 subscale and 0.75 for the modified S3 subscale.
   In 2009, Cramer et al. conducted an experiment in which                          This result shows that the English adaptation of the NARS
they examined how people perceived a robot when physical                            S1 subscale can be used with people who speak English.
contact and help style were factors [3]. In the between sub-                        Further, people are able to provide consistent ratings for the
jects online survey, participants were asked to watch one for                       NARS scale for interactions with robots shown in videos, as
four videos featuring a Robosapien helping a woman using a                          opposed to in person.
word processing application on a computer. The woman has
a computer problem, and the robot gives advice as to what
she should to to recover her work. In the condition with
                                                                                    3.   CASE STUDY: TELEPRESENCE ROBOTS
physical contact, the woman tapped the Robosapien to get                               The Robosapien actor in the videos was intended to be
its attention for help; the Robosapien touched the woman                            perceived as an autonomous, independent agent. Our goal
                                                                                    is to determine if NARS can also be used when the robot
                                                                                    is known to be controlled by a human. Using Amazon’s
                                                                                    Mechanical Turk, we conducted an online video survey of
Permission to make digital or hard copies of all or part of this work for           five telepresence robots in a between subjects experiment
personal or classroom use is granted without fee provided that copies are           with 80 participants.
not made or distributed for profit or commercial advantage and that copies              We used a slightly modified version of the NARS scale
bear this notice and the full citation on the first page. To copy otherwise, to      presented in English in [7]. For our questions, we inserted
republish, to post on servers or to redistribute to lists, requires prior specific   one duplicate question in the S2 subscale. The duplicate
permission and/or a fee.
PerMIS ’10, September 28-30, 2010, Baltimore, MD, USA                               question rephrased “I feel that in the future society will be
Copyright 20XX ACM X-XXXXX-XX-X/XX/XX ...$10.00.                                    dominated by robots” to “I feel that in the future, robots
                       Table 1: A sampling of studies using NARS as a performance measure
 Study                 Year   n     Language       Robot         Study Description
 Nomura et al. [6]     2004   240   Japanese       Robovie       Investigated the use of psychological scales for use in HRI
 Nomura et al. [5]     2006    53   Japanese       Robovie       Investigated the relationship between peoples’ negative
                                                                 attitudes and their interaction with robots
 Woods et al. [9]      2006   159    English    robot images     Examines robots from children’s perspective
 Bartneck et al. [2]   2007   467    English        Aibo         Investigated effects of cultural differences and prior robot experience
 Nomura et al. [7]     2007    17   Japanese     Robovie-M       Investigated relationship between peoples’ attitudes
                                                                 towards and the distance maintained during interaction
 Riek et al. [8]       2010   16     English       BERTI         Examine videos of humanoid gesture (beckon, give, and handshake)

                                                                     positively to our question modification. In the baseline NARS
Table 2: Chronbach alpha values of NARS subscales                    rating, “commonplace” averaged 3.96 (SD=0.76) and “dom-
               Base- After Robo- After telepresence
                                                                     inate” averaged 2.73 (SD=1.25) (p<0.01, t(69)=7.45). Af-
                line sapien video   robots videos
                                                                     ter seeing the Robosapien video, “commonplace” averaged
      S1       0.76      0.84           0.87                         3.87 (SD=0.87) and “dominate” averaged 2.85 (SD=1.35)
 S2 (original) 0.76      0.81           0.87                         (p<0.01, t(69)=5.42). After seeing the telepresence robot
 S2 (modified)  0.69      0.74           0.86                         videos, “commonplace” averaged 3.93 (SD=0.99) and “dom-
      S3       0.69      0.74           0.83                         inate” averaged 2.87 (SD=1.35) (p<0.01, t(139)=7.63).
                                                                        Current Work. We found that overall, NARS can be
will be commonplace in society.” As per the original Likert          used as a performance measure for videos of robots, which
scale from [7], participants were asked to rate the NARS             is consistent with prior works (e.g., [3, 8]). Further, we have
statements on a scale from 1 (I strongly disagree) to 5 (I           shown that NARS can be used with videos of robots which
strongly agree) with 3 as “undecided.”                               are known to be operated by people. We are currently run-
   Experimental Design. In the survey, participants were             ning an experiment to validate NARS with people operating
asked to provide their demographic information, including            telepresence robots in person. The results of this experiment
age, gender, occupation, country of citizenship, and pet own-        will be included in the full paper.
ership. They then completed a baseline NARS rating. We
also solicited information about their prior robot experience        4.   REFERENCES
and their video game usage in the last 12 months.                    [1] C. Bartneck, T. Nomura, T. Kanda, T. Suzuki, and
   Participants were asked to view three videos, and answer              K. Kennsuke. A Cross-Cultural Study on Attitudes Towards
the NARS questions after each video. In the first video, par-             Robots. In Proc. of the HCI Intl., 2005.
ticipants saw one of four of the Robosapien videos (chosen at        [2] C. Bartneck, T. Suzuki, T. Kanda, and T. Nomura. The
random) used in Cramer et al.’s study [3]. The second video              Influence of People’s Culture and Prior Experiences with
was of the Anybots’ QB robot. The remaining video showed                 Aibo on Their Attitude Towards Robots. AI & Society,
one of four telepresence robots: Willow Garage’s Texai, Vgo-             21(1):217–230, 2007.
Com’s Vgo, RoboDynamics’ TiLR, or SuperDroid’s RP2W.                 [3] H. Cramer, N. Kemper, A. Amin, V. Evers, and B. Wielinga.
We created four versions of the survey for each of the re-               “Give Me a Hug”: The Effects of Touch and Autonomy on
                                                                         People’s Responses to Embodied Social Agents. Computer
maining telepresence robots; twenty participants saw each
                                                                         Animation and Virtual Worlds, pages 437–445, 2009.
video. We randomized whether the participants saw the QB
                                                                     [4] L. Cronbach. Coefficient Alpha and the Internal Structure of
robot as the second or third video.
                                                                         Tests. Psychometrika, 16(3):297–334, 1951.
   Results. Seventy participants provided usable data. We
                                                                     [5] T. Nomura, T. Kanda, and T. Suzuki. Experimental
discarded 10 participants’ data due to submission of dupli-
                                                                         Investigation into Influence of Negative Attitudes toward
cate surveys, exiting the survey before completion, or in-
                                                                         Robots on Human-robot Interaction. AI & Society,
correctly answering validation questions. Participants’ ages
                                                                         20(2):138–150, 2006.
ranged from 18 to 59 years old with the average age of
                                                                     [6] T. Nomura, T. Kanda, T. Suzuki, and K. Kato. Psychology
28.66 (SD=8.73); forty participants were male and twenty-                in Human-robot Communication: An Attempt through
eight female. Twenty-nine participants reported the United               Investigation of Negative Attitudes and Anxiety toward
States as their country of citizenship. Thirty-one partici-              Robots. In Proc. the 13th IEEE Intl. Workshop on Robot
pants were citizens of India, and the remaining were citizens            and Human Interactive Communication, pages 35–40, 2004.
of Australia, Canada, Greece, Honduras, Russia, Singapore,           [7] T. Nomura, T. Shintani, K. Fujii, and K. Hokabe.
Switzerland, and United Kingdom.                                         Experimental Investigation of Relationships between
   For all seventy responses, we computed the Chronbach al-              Anxiety, Negative Attitudes, and Allowable Distance of
pha values for the baseline NARS response the participants               Robots. In Proc. of the 2nd IASTED Intl. Conf. on Human
saw in the demographic information section, for the replica-             Computer Interaction, pages 13–18. ACTA Press, 2007.
tion of the Robosapien videos, and for all of the telepresence       [8] L. Riek, T. Rabinowitch, P. Bremner, A. Pipe, M. Fraser,
robots in aggregate (shown in Table 2). In the full paper,               and P. Robinson. Cooperative Gestures: Effective Signaling
we will compute the Chronbach alpha values for each of the               for Humanoid Robots. In Proc. of the 5th ACM/IEEE Intl.
NARS subscales in more detail based on demographic infor-                Conf. on Human-robot Interaction, pages 61–68, 2010.
mation (e.g., for participants from only the US, participants        [9] S. Woods. Exploring the Design Space of Robots: Children’s
not from the US but who speak English, gender).                          Perspectives. Interacting with Computers, 18(6):1390–1418,
   We found that participants responded significantly more                2006.

To top