The ACCEPTABILITY of Caregiver Robots in Elderly People

Melissa Ferretti, Giovanna Morgavi and Gianmarco Veruggio

IEIIT (Istituto di Elettronica e Ingegneria dell’Informazione e delle Telecomunicazioni), Torino, Italy

CNR (Consiglio Nazionale delle Ricerche), Via De Marini 6,16149 Genoa, Italy

Keywords: Caregiver Robots, Robots for Elderly People, Robot Acceptability.

Abstract: During the last few years, due to the aging of the population, many scientists have developed ICT tools to

offer elderly people an independent life at home as long as possible. Most of these researchers focused their

efforts on problem solving without adequate care to the agreeability and/or the acceptability of these ICT

objects for their users. These resulting artifacts will hardly be used in real life by the users for which they

have been developed. In this paper, we will present an experiment done on 202 over 65 elderly people on

the acceptability and the likeness features a caregiver robot must have. From the classification and analysis

of the emotions elicited by the physical/appearance characteristics of 25 different real robot pictures we

found some interesting results for appealing or unpleasant features for caregiver robot design.

1 INTRODUCTION

Populations around the world are rapidly aging.

According to an estimation by the OECD by the

middle of the 21st century the number of older

people will exceed 2 billion (around 21% of the

world's population), and this trend will affect and

cover not only industrialized nations but also

developing nations (OECD, 2015).

To cope with this growing aging population,

societies will need to adapt to this changing

demographic and invest in healthy aging, enabling

individuals to live both longer and healthier lives.

Finding a way to create and strengthen

conditions for an "active aging", which also aims to

maintain the independence at home of the elderly

population, can be a serious challenge but also a

great opportunity.

Technology, and particularly AI, could be part of

the solution to this problem, offering support for

older adults with the difficulties and challenges

associated with aging (Pollack, 2005). Specifically

robots could have great potential for providing

assistance to older adults in their own homes and so

the question about robot acceptance is particularly

relevant for proper artifact design.

Various researches focused on the study of the

functions that a caregiver robot should perform.

Numerous attempts to create robotic tools, both in

development and commercialization, have been

create to carry out specific tasks to help the elderly

live at home for longer by performing activities such

as medication management, house keeping, social

entertainemen and providing emergency monitoring.

However, as shown in literature, technology

applications developed for senior users are often

discarded due to factors that are specific to this age

group of people. Acceptance of a robotic caregiver is

a complex and multifaceted issue. Studies conducted

on elderly people is usage of ICT tools showed how

the reluctance to adopt new technological

instruments is not only due to a lack of skills but,

also, to the lack of perception of advantages and

benefits of using these tools. To ensure acceptance

of these new technological tools the age-related

changes in perceptual, motor and cognitive abilities

must be considered. Combined with these really key

aspects, it is necessary to recognize the importance

of the compensatory process that older people

develop to adapt to their changes and to understand

the crucial role played by motivation, affection, and

experience in every social interaction. In this

context, if we want to increase the likelihood that

people will utilize robot assistance, acceptance is a

key factor. Indeed, if the development of these

robots designed to solve pretended problems, does

not lead to agreeable and/or acceptable objects to the

elderly, they will hardly be used.

Ferretti, M., Morgavi, G. and Veruggio, G.

The ACCEPTABILITY of Caregiver Robots in Elderly People.

DOI: 10.5220/0006674301110118

In Proceedings of the 4th International Conference on Information and Communication Technologies for Ageing Well and e-Health (ICT4AWE 2018), pages 111-118

ISBN: 978-989-758-299-8

111

As a result, we decided to focalize our attention

on older adults’ attitudes and preferences for robots,

focusing on the aspects that are not functional but

kinesthetic, because the acceptability of these tools,

for this age group, depends heavily on empathetic

factors. Keeping this in mind, we could be able to

design robot capable to serve the needs of the

elderly.

2 CAREGIVER ROBOT

This paper is part of the extensive research

landscape that is being carried out today in the field

of social robotics. Researches in eldercare proposed

robots to be a form of assistive technology with a

great potential to support older adults, to maintain

their independence, and to enhance their well-being

(Ezer, et al., 2009).

In literature, assistive robots are classified in two

groups according to the function for which they

were developed: rehabilitation robots and social

robots (Broekens, et al., 2009).

Social robots, used in eldercare studies, can then

be divided into two other categories: service type

robots, developed to be used as assistive devices,

and companion type robots, developed to enhance

health and psychological wellbeing.

The research in this field is rich and fervid and

the technology development in the homecare robotic

field is developing faster and faster. Probably in the

near future, robot caregivers will become feasible

and affordable, but, currently, this technology

development is mostly technology driven. The

question if the elderly would accept a robotic

assistant at home has still to be more deeply

investigated.

In literature, most of the studies measured the

acceptance of specific robots with limited

functionality (Smarr, et al., 2013). Some papers

cover the definition of the tasks that elderly could

delegate to robot assistant. In (Ziefle and Calero,

2017) for example, particular situations where

elderly people can accept that some tasks are

performed by a robot on behalf of humans are

discussed. But this gives little information about

general attitudes and perceptions of the elderly about

robots because it is too related to the contingency of

the performing task.

Other studies investigated the relationship

between appearance and functionalities, stated that

appearence influences the assumptions that people

make of a robot and of the tasks correlated to it

(Goetz, et al., 2003). In this meaning, appearance

must support the real expectations of the robot's

skills. The more the user gets a clear idea of what

the machine can do, the less he will be disappointed

when using it (Kaplan, 2005). Within this vision,

functionalities of the robot loose weight and the

appearance should be designed just to help users

build a mental model of the robot usage (Lohse, et

al., 2008).

On the other hand, researches also emphasized

how the technologies for assistance, designed to

facilitate autonomy, are often perceived as a

handicap or aging signal and this realization can lead

to their rejection. Therefore the design of assistive

ICT tools should be universal. It should aim at de-

stigmatizing assistive robots making them appealing

and useful for everyone and not just for the elderly

or disabled (Wu , et al., 2014).

Finally, as highlighted by Van der Heijden, in

‘hedonic systems’, the concept of enjoyment is

crucial for the intention to use a techological tool

(Van der Heijden, 2004). Obviously, in eldercare,

we can’t say that a robot is developed just for

entertaining, but enjoyment needs to be part of the

acceptance model for robotic technology.

Our research moves right from this assumption

and seeks to understand in advance what the

physical characteristics are that affect acceptability

and enjoyability, making them the basis for future

developments and functional studies.

3 RESEARCH QUESTION AND

PURPOSE OF THE STUDY

This paper examines the physical features that make

a caregiver robot fit and usable in order to

understand the peculiarities such device should have

to be really used by the elderly at home. The

caregiver robot should increase independent living

and social participation of older people in relatively

good health, comfort and safety.

In our experiment, we investigate what

appearance a robot shoud have. Our analysis

reflects on physical aspects of the robot rather than

on functional aspects. We designed an experiment

to try to identify empathetic features that, in some

way, facilitate the acceptance and desirability of the

robot by the elderly.

This experiment was conducted on 202 italian

people aged over 65.

Table 1 shows the robots we selected to be

evaluated within this experiment. These robots have

been chosen among various artifacts developed in

ICT4AWE 2018 - 4th International Conference on Information and Communication Technologies for Ageing Well and e-Health

112

the world research scene. We did not limit our

choice among social assistive robots, but we also

took into account machines belonging to different

fields of application like Kismet or ICube.

We created twenty five cards, one for each

selected robot, to highlight the physical and

functional characteristics and robot dimensions.

Each card contains two or more color images of

one robot with elements (people or objects of known

size) that allowed the observer to informally infer on

dimensions and functions of the robot.

Figure 1: Experiment card example. As you can see,

picture 1 displays the Nao robot dimension and its

possible social interactions. Picture 2 displays the

entertainment activity of Nao robot.

For our experiment we chose to use a robot

classification that can be partially riconduced to the

Brokens et al. paper (Broekens, et al., 2009). The

pool of tested robots was composed as shown in

Table 1.

The first group is composed of

medical/rehabilitation robots. In this category, the

enphasis is focused on the physical assistive

technology and function (i.e. Riba II, a robot

developed to perform patient-transfer tasks

(Toshiharu, et al., 2010)).

The second group is representative of the social

robots, systems that can be perceived as social

entities with communication capacities. In this case,

as stated in literature, we complied with the

distinction between service robots and social robots.

Service type robots typically investigate which

social features can lead to the acceptance of a

robotic device at home and how these same social

features can facilitate the actual use of the device.

Examples of these researches are the German Care-

o-bot, a robotic assistant that supports people in their

daily living at home performing common tasks like

offering drinks, setting the table, switching on the

TV or the radio and even calling for rescue service

in case of emergency (Graf, et al., 2004), or

Giraffplus, a robot developed to check elderly

health, ready to rescue in case of emergency and

able to put users’video calls through to their

relatives and physicians (Coradeschi, et al., 2014).

Companion type robots focus on pet-like

companionship, like the Japanese seal-shaped robot,

Paro, (Wada, et al., 2003 and Shibata and Wada,

2011), the Sony small robot dog, Aibo, or the

robotic Japanese cat, Yume Neko Venus.

Table 1: List of the 25 robots evaluated within the

experiment.

Medical/rehabilitation

robots

iRobi Q

Riba II

Medical robot

Social robots

Companion

type

Aibo

Yume Neko

Venus

NAO

Paro

Service

type

Roomba

-O-Bot

Giraff

lus

Asimo

Electronic

Sourveillance

Turtle Bot

Romeo

Chess

Terminato

Ramci

PR2

General purpose

robots

CB2

ICube

Kismet

Mathilda

Albert Hubo

Wall-E

Kobian

Finally, we added the general purpose robots group

where we put robots that are not classifiable within

the two previous groups. They don’t have a specific

The ACCEPTABILITY of Caregiver Robots in Elderly People

113

function clearly understandable by looking at the

pictures.

3.1 Method

3.1.1 The Sample

This experiment was conducted on 202 italian

people aged over 65, participation was voluntary and

anonimity was guaranteed. Each participant signed a

disclaimer sheet for privacy. Data was collected

through personal interviews conducted by graduates

in psychology. The duration of each experiment

session was approximately 1 hour.

The experiment started by collecting information

about participants’ demographics (age, gender,

profession and education).

Figure 2: Sample distribution age by gender.

The response sample was composed of elderly

Italian adults living independently (N = 202), aged

65 to 87 (M = 74 years; SD =5,5 years). 59% of the

sample was composed by female and 41% by male.

Participants varied in their educational

background, with 39% having college or university

education and with 61% having less than a formal

college education (35% having only a first grade

education).

3.1.2 The Experiment Process

Each participant was asked to judge the acceptability

of the robot based on the feeling elicited by the

observation of each card containing the picture of

the robot and to put the cards in order by preference:

first the preferred one and last the less liked.

The conductor of the experiment, to facilitate the

carrying out of this task, presented the participant

Figure 3: Participant education distribution.

with cards in pairs. Then, he/she asked the question:

‘Which robot among these two would you prefer to

have at home?’. Among these two cards, the

participant had to choose which one he liked more.

Iterating this process, all the 25 cards were sorted in

order of preference.

No verbal information on the role and/or

function of the robot was given to the participants.

Conductors were instructed, if questioned about the

robot, not to give direct answers, but to stimulate

reflection by letting the participants think what

he/she might infer from the images.

4 RESULTS

A preliminary analysis was performed in order to

evaluate only cards classified in the first or in the

last position. Figure 4 shows the number of times

each robot obtained the first position.

In the right space of figure 5, the five robots that

got the highest number of first places are shown:

they are Aibo (12%), the little Sony robot dog,

Yume Neko Venus (11%), the Japanese robotic cat,

ICube (9%), the baby-like robot developed by IIT,

Paro (8%), the small Japanese seal-shaped robot and

Giraffplus (7%) the social communication robot.

The experiment results showed a strong

preference (about 40% of the sample) for robots

similar to small animals or babies.

By analyzing the distribution of the score of

robots with the last ratings, we can observe that the

worst classified robot is CB2, the baby-like robot

(17% of the participants placed it in the last

position). CB2 and ICube are both baby robots: what

is the difference between them that makes such a big

difference in the preferences? CB2 is bigger then a

human being, ICube is smaller. CB2 has a more

ICT4AWE 2018 - 4th International Conference on Information and Communication Technologies for Ageing Well and e-Health

114

Figure 4: Distribution of number of the first position

scores for each robot. On the right side of the picture, an

image of the 5 first classified robots is shown.

detailed face than iCube, in this case confirming the

Uncanney Valley theory (Mori, 1970). And last , but

not least, in recent times, ICube robots have been

presented many Italian TV shows and

advertisements and maybe its look became familiar.

Figure 5: Distribution of number of the last position scores

for each robot. On the right side of the picture, an image of

the 5 worst classified robots is shown.

It is interesting to underline that the social robots

Kismet, Albert Hubo, Kobian and Asimo together

account for 46% of the last position. What do they

have in common? All of them show a human-like

appearance and all of them have dimensions that are

greater or equal to human dimensions.

This analysis, however, gives us only a partial

picture of the results of the experiment and

therefore, in order to be able to take into account the

intermediate positions, we made an overall analysis

of the order of classification of the cards by

Table 2: Mean (M), Standard deviation (SD), Median

(Me) and Mode (Mo) of the robot scores, ordered by

Median.

Robot M SD Me Mo

1 GiraffPlus 9,59 7,04 8 2

2 Aibo 9,93 6,7 9 1

3 Roomba 10,98 6,55 10 4

4 NAO 10,48 6,32 10 5

5 Paro 10,94 7,87 10 2

6 Asimo 11,79 7,01 10,5 6

7 Yume Neko Venus 11,25 7,96 10,5 1

8 Romeo 11,68 6,92 11 13

9 I Robi Q 11,79 7,02 12 2

10 Turtlebot 13,02 6,36 12 7

11 Ca

-O-Bot 12,83 6,94 13 18

12 Riba II 12,48 7,83 13 2

13 ICube 12,19 7,4 13 1

14 Wall-E 13,38 6,53 13,5 4

15 Pe

13,43 7,03 14 22

16 Medical 13,66 6,71 14 21

17 Chess Terminato

13,29 6,36 14 14

18 PR2 14,82 5,98 14 12

19 Ramcip 14,43 6,07 14,5 11

20 CB2 16 7,34 16 25

21 Kismet 15,35 7,29 16 21

22 Mathilda 13,96 7,91 16 23

23 Sourveillance 15,24 6,53 17 22

24 Kobian 16,2 6,78 17,5 15

25 Albert Hubo 16,3 7,21 18,5 24

extrapolating average, median and mode from the

data sample. As seen in Table 2, the arrival order

varies according to the statistical value considered.

If we take the median as a significant value in the

first five positions, we find Giraffplus, Aibo,

Roomba, Paro and Nao. Among them, there are

Paro, Aibo and GiraffPlus, which also appeared

among the top 5 of figure 4.

By considering the median, more than 50% of

the sample liked Giraffplus and put the robot within

the top 8 positions.

Even if Giraffplus is taller than human beings, it

is probably non considered dangerous since it is very

thin and its functions (allowing video

communication with other people), coupled with a

non similarity to human being, contribute to rating it

in a good position for a wide number of people.

Second, we find Aibo. For 50% of the

participants, Aibo is rated between the first and the

ninth position. In third place we find Nao, Paro and

Roomba (the cleaning robot). For more than 50% of

the sample, their rating is located within the top 10

rates.

The ACCEPTABILITY of Caregiver Robots in Elderly People

115

Figure 6: Distribution of Mean, Median and Mode ordered

by Median-Mean of robot rating.

To measure the statistical dispersion, we divided the

data set into quartile and we computed the

interquartile range

(IQR) that is equal to the

difference between third and first

quartiles. Quartiles

divide a rank-ordered data set into four equal parts.

The values that separate parts are called the first,

second, and third quartiles; and they are denoted by

Q1, Q2, and Q3, respectively.

For each robot card, figure 7 shows the median

of the position value with IQR didtribution.

Figure 7: IQR distribution for median of the position

values of the card position for each robot.

The interquartile gap indicates a measure of how

many values deviate from the sample median.

If we consider the variability of the first 5

robots with lower median ( i.e. the 5 best classified

robots by using the median of the position rate as

evaluation parameter) we can observe that while

Giraffplus, Aibo, Nao and Roomba show a

comparable variability, Paro show a larger

variability in the position.

Giraffplus, which is the best classified in this

case, shows IQR valus larger then Aibo, Nao and

not perfectly centered.

Figure 8: Giraffplus position rate distribution. Q

and Q

are the first and the third Quartile respectively and M

the median value.

Indeed, the 50% of rates falls between Q

and Q

but

it is shifted to Q

Figure 9: Position rate distribution of Aibo, Nao, Roomba

and Paro.

Figure 9 shows the position rate distribution of Aibo,

NAO , Roomba and Paro. The high IQR of Paro

indicates that this robot shows some characteristics

that are considered positive for some and for others

are neutral or negative. Paro is a social robot that

‘asks for caregiving’ and its liking may depend on

the emotional features, on the story or on the mood

the evaluator is experiencing.

ICT4AWE 2018 - 4th International Conference on Information and Communication Technologies for Ageing Well and e-Health

116

Figure 10: Albert Hubo position rate distribution. Q

and

are the first and the third Quartile respectively and M

is the median value.

Robots rated with the 5 highest median value in the

card evaluation process mostly confirm the previous

deductions. Albert Hubo, the Japanese robot with the

Albert Einstein face, is placed in the worst position,

followed by Kobian, another Japanese humanoid

robot with human dimensions that can display seven

different emotions. The forth worst position is

occupied by CB2, the Japanese baby robot, and in

the fifth worst position we can find Kismet, the MIT

robot face able to recognize and reproduce emotions.

The presence of humanoid robots in the worst

positions also confirms also for elderly people the

theory of the Uncanny Valley proposed by Masahiro

Mori (Mori, 1970). Indeed humanoid objects which

appear almost, but not exactly, like real human

beings elicit uncanny or feelings or feelings of

strangeness and revulsion in observers. Moreover,

the dimension of these humanoid robots seems to be

critical to worsen the feeling of discomfort caused

by the humanoid robots. The surveillance robot can

affect everyone’s need of privacy. A robot

performing surveillance can be explicitly perceived

as a prosthesis, a privacy intrusion or a signal of loss

of independence and autonomy.

Furthermore, the presence in the picture of the

controller tablet induced a feeling of technological

inability.

5 CONCLUSIONS

In this paper, we presented an experiment on the

acceptability of robot caregivers done with 202

elderly people as participants. Preliminary results

suggest some important tips for designing a usable

artefact. While most critical negative factors are

Figure 11: Position rate distribution of Kobian, the

sorveillace robot, CB2 and Kismet.

large sizes, excess of human similarity, the feeling

of low level of controllability or an overly

mechanical aspect. The most popular robots seem to

be the ones that in some way maintain their robot

likeness. They should be small and can be perceived

as a toy or a puppy. Even if the puppy likeness

seems to elicit empathy, closeness, and confidence,

the resemblance to human babies seems not

sufficient to guarantee appeal. The robot should

maintain its robot identity, clearly recognizable. This

experiment showed that the most important features

therefore seem to be small sizes, cartoon traits

and/or animal appearances.

Naturally, this suggestion is critical because it is

difficult or impossible for small robots to perform

some service tasks. Some solutions can, probably, be

found in the direction of the distribution of services:

many small robots performing different tasks. Other

solutions can be reached by involving elderly people

in new robot design.

Last but not least, a robot caregiver should help

elderly people, but should also facilitate

communication with other human beings ( as the

high rate of Giraffplus shows).

Our results advance the understanding of older

adults’ attitudes and preferences which may

influence the design of robots more likely to be

accepted by older adults.

Future research will investigate in detail the feeling

elicited by the single robot, how older adults interact

with a physical robot and how/if attitudes change

over time.

The ACCEPTABILITY of Caregiver Robots in Elderly People

117

REFERENCES

Broekens, J., Heerink, M., Rosendal, H., 2009. Assistive

social robots in elderly care: a

review. Gerontechnology Volume 8(2):94–103.

Coradeschi S. et al., 2014. GiraffPlus: A System for

Monitoring Activities and Physiological Parameters

and Promoting Social Interaction for Elderly. In:

Hippe Z., Kulikowski J., Mroczek T., Wtorek J. (eds)

Human-Computer Systems Interaction: Backgrounds

and Applications 3. Advances in Intelligent Systems

and Computing, vol 300. Springer, Cham.

Ezer, N., Fisk, AD., Rogers, WA., 2009. More than a

servant: self-reported willingness of younger and

older adults to having a robot perform interactive and

critical tasks in the home. Poster presented at: 53rd

Annual Meeting of Human Factors and Ergonomics

Society, San Antonio, TX.

Goetz, J., Kiesler, S., Powers, A., 2003. Matching robot

appearance and behaviour to tasks to improve human-

robot cooperation. In: Proceedings of the 12

IEIIT

Workshop on Robot and Human Interactive

Communication (ROMAN 2003), San Francisco, CA,

pp. 55-60.

Graf, B., Hans, M., Schraft, R.D., 2004. Care-O-bot II—

Development of a Next Generation Robotic Home

Assistant. Autonomous Robots (2004) 16: 193.

Kaplan, F., 2005. Everyday robotics: robots as everyday

objects. In: Proceedings of Soc-Eusai 2005, Grenoble,

France, pp. 59-64.

Lohse, M., Hegel, F., Wrede, B., 2008. Domestic

applications for social robots: an online survey on the

influence of appearance and capabilities. Journal of

Physical Agents. Vol. 2, No. 2. ISSN 1888-0258, pp.

21-32.

Mori, M., 1970. Bukimi no tani. (The Uncanny Valley,

trans. by MacDorman, K.F. and Kageki, N. under

authorization by Masahiro Mori. IEEE Robotics &

Automation Magazine, 98–100.). Energy, 7(4) pp. 33–

35, (Originally in Japanese).

OECD, 2015. Ageing: Debate the Issues. OECD Insights,

OECD Publishing, Paris.

Pollack, M., 2005. Intelligent technology for an aging

population: the use of AI to assist elders with cognitive

impairment. AI Magazine Volume 26 Number 2.

Shibata, T., Wada, K., 2011. Robot Therapy: A New

Approach for Mental Healthcare of the Elderly – A

Mini-Review. Gerontology; 57:378–386.

Smarr, C., Mitzner, T., Beer, J., Prakash, A., Chen, T.,

Kemp, C., Rogers, W., 2014. Domestic Robots for

Older Adults: Attitudes, Preferences, and Potential.

Int J Soc Robot 6:229–247.

Toshiharu, M., Shinya, H., Hiromichi, N., Yo, K., Yuki,

S., Shijie, G., Shigeyuki, H., 2010, Development of a

Nursing-Care Assistant Robot RIBA That Can Lift a

Human in Its Arms. The 2010 IEEE/RSJ International

Conference on Intelligent Robots and Systems

October 18-22, 2010, Taipei, Taiwan.

Van der Heijden, H., 2004. Users acceptance of hedonic

information systems. Manag Inf Syst Q 28(4):695-704.

Wada, K., Shibata, T., Saito, T., Tanie, K., 2003. Effects of

robot assisted activity to elderly people who stay at a

health service facility for the aged. In: Proceedings of

IROS 2003: The IEEE/RSJ International Conference

on Intelligent Robots and Systems. pp 2847-2852.

Wu, Y., Wrobel, J., Cornuet, M., Kerhervé, H., Damnée,

S., Rigaud, A. 2014. Acceptance of an assistive robot

in older adults: a mixed-method study of human–robot

interaction over a 1-month period in the Living Lab

setting. Clinical Interventions in Aging 2014:9 801–

811.

Ziefle, M., Calero, V.A., 2017. Domestic Robots for

Homecare: A Technology Acceptance Perspective. In:

Zhou J., Salvendy G. (eds) Human Aspects of IT for

the Aged Population. Aging, Design and User

Experience. ITAP 2017. Lecture Notes in Computer

Science, vol 10297. Springer, Cham.

ICT4AWE 2018 - 4th International Conference on Information and Communication Technologies for Ageing Well and e-Health

118