Hybrid Approach to Promote Social Interaction with Children with

Autism Spectrum Disorder

Vinícius Silva

, Filomena Soares

, João Sena Esteves

, Ana P. Pereira

, Celina P. Leão

and Sandra Queirós

Centro Algoritmi, Engineering School, University of Minho, 4800-058 Guimarães, Portugal

Research Centre on Education, Institute of Education, University of Minho, Braga, Portugal

sandraqueiros94@gmail.com

Keywords: Human Robot Interaction, Autism Spectrum Disorder, Playware Technology.

Abstract: The comprehension of the emotional state of others is paramount for a successful human interaction.

Individuals with Autism Spectrum Disorder (ASD) have impairments in social communication and,

consequently, they have difficulties to interpret others’ state of mind. In order to tackle this issue, researchers

have been proposing the use of technological solutions to assist children with ASD, particularly in imitation

and emotion recognition tasks. Social robots and Objects with Playware Technology (OPT) have been

employed as intervention tools with children with ASD. This work presents an approach combining both

technologies (robots and OPT), in a hybrid way, with the goal of promoting social interaction with children

with ASD. Moreover, a new OPT device was developed to be used as an add-on to the human-robot interaction

with children with ASD in two emotion recognition tasks – recognize and storytelling. A pilot study was

conducted with children with ASD to evaluate the proposed method. All children successfully participated in

the activities. Moreover, children significantly gazed longer towards the OPT during the storytelling scenario

as the OPT device displayed visual cues, supporting that using a visual cue may be fundamental in helping

children with ASD understand requests and tasks.

1 INTRODUCTION

Autism Spectrum disorder (ASD) is a

neurodevelopment disability that affects 1 in 54

individuals. It is characterized by the diagnostic

criteria that include impairments in social

communication and social interaction, with the

existence of restricted, repetitive patterns of

behaviour, or activities that may continue throughout

life (American Psychiatric Association, 2013). The

diagnosis can be done correctly in early stages of life

(around the 36 months of age). Due to the diversity

and specificities of symptoms, developing effective

intervention is still challenging.

In order to tackle this issue, new forms of

intervention have been explored and conducted in the

last years by employing the use of technological

https://orcid.org/0000-0003-0082-343X

https://orcid.org/0000-0002-4438-6713

https://orcid.org/0000-0002-3492-1786

https://orcid.org/0000-0003-4611-7602

https://orcid.org/0000-0003-3725-5771

devices such as robots, tangible interfaces/Objects

with Playware Technology (OPT) and mechanical

components, among others. Indeed, some studies

(Dautenhahn & Werry, 2004; Tapus et al., 2012)

conducted by using technological tools showed that

children with ASD have great affinity with them. In

particular, it has been shown that individuals

demonstrated improvements in social behaviours

such as imitation (Fujimoto, Matsumoto, de Silva,

Kobayashi, & Higashi, 2011), eye gaze, and motor

ability, while increasing attention (Kim, Paul, Shic, &

Scassellati, 2012) when interacting with robots.

Moreover, it was also identified that children with

ASD may exhibit certain positive social behaviours

when interacting with robots in contrast to what is

perceived when interacting with their peers,

caregivers, and professionals (Gillesen, Barakova,

Silva, V., Soares, F., Esteves, J., Pereira, A., Leão, C. and Queirós, S.

Hybrid Approach to Promote Social Interaction with Children with Autism Spectrum Disorder.

DOI: 10.5220/0010468600690077

In Proceedings of the 7th International Conference on Information and Communication Technologies for Ageing Well and e-Health (ICT4AWE 2021), pages 69-77

ISBN: 978-989-758-506-7

Huskens, & Feijs, 2011). In addition to being

repeatable and objective, social robots are designed to

identify, measure, and react to social behaviours,

offering an exceptional occasion for quantifying

social behaviours (Tapus, Member, & Scassellati,

2007). Therefore, robots may be very promising on

intervention settings with children with autism,

especially in tasks such as emotion recognition and

collaborative peers’ interaction.

Several physical designs, ranging from simple

designs such as four-wheeled mobile robots (Ferrari,

Robins, & Dautenhahn, 2009) to many levels of

anthropomorphic forms, including humanoid (Soares

et al., 2019), animal-like (Breazeal, 2000), and

machine-like systems (Michaud et al., 2005), have

been used. Since the robot’s physical appearance

plays an important role in the interaction process with

a person, recent research in the area of social robots

have been consistently using robots with a humanoid-

like design (Pennisi et al., 2016), especially in tasks

of imitation and emotion recognition, offering a great

potential for generalisation. The FACE (Mazzei et al.,

2011) and ZECA (Soares et al., 2019) projects use

facial expressive humanoid robots in emotion

recognition tasks to promote social interaction with

individuals with ASD.

Analogous to the use of social robots, OPT

devices have also been used as a form of interaction

with children with ASD. With the goal of offering

playful experiences to the end user, these devices are

tangible interfaces developed for children’s play

(Lund, Klitbo, & Jessen, 2005). The term “playware”

is suggested as a combination of intelligent hardware

and software, emphasizing the role of interplay

between morphology and control using processing,

input, and output.

Few works focusing on OPT with different

configurations such as modular buttons, coloured

puzzle tiles (Lund & Marti, 2009), Lego-like building

blocks (Barajas, Al Osman, & Shirmohammadi,

2017), interactive screens (Boucenna et al., 2014),

among others have been proposed. An example of a

work where OPT devices were used with children

with ASD consisted in designing interactive tiles as a

modular robotic playware with the goal of being

flexible in both setup and activity building for the end

user, allowing easy creation of games (Lund, Dam

Pedersen, & Beck, 2009). The tiles had a quadratic

shape with self-contained energy source and wireless

communication (local and global), and different

games. An example of a game conducted during

experiments with children with ASD consisted in

mixing the tiles in different combinations to produce

more colours. More specifically, there were three

main tiles with fixed colours (mainly red, green, and

blue), and using the secondary tiles (a total of 12),

with the property of changing their colours

accordingly to their neighbours colour, a new colour

could be created by mixing the neighbours tiles

colours. For example, if a secondary tile was placed

between two main tiles of colours red and blue, the

middle tile (secondary) would change its colour to

purple, blending the two main colours. The

experiments carried out with seven children with

ASD allowed the authors to conclude that devices

built with playware technology may offer an

interesting novel research direction. Through this

research direction, this issue will be further

investigated in order to verify how playware can be

used as playful tool for cognitive challenged children,

giving them a playful experience and automatically

infer the playful interaction to provide insight (and

possible a diagnosis).

Following this trend, it was proposed and

evaluated a novel approach combining both

technologies (robots and OPT), in a hybrid way, with

the goal of promoting social interaction with children

with ASD (Silva, Soares, Esteves, & Pereira, 2018).

The present work shows the developments of this

approach as well as the development of a new OPT

device to be used as an add-on to the human-robot

interaction with children with ASD in emotion

recognition tasks. This hybrid approach was

evaluated through experiments conducted with these

children. The main goal of the pilot study was to

assess the constraints of both the game design as well

as the OPT rather than to evaluate the performance of

each child.

The rest of this paper is organized as follows:

Section 2 presents the proposed approach with the

description of the OPT; Section 3 shows the results;

Section 4 discusses the results obtained; the conclus-

ions and future work are addressed in Section 5.

2 MATERIAL AND METHODS

This section provides the description of the developed

OPT with the procedures followed to evaluate the

current proposed approach.

2.1 Proposed System

The system, depicted in Fig. 1, consists of a facial

expressive humanoid robot, a processing unit, and the

OPT device PlayBrick.

The humanoid robot used is the model Zeno R50

from Robokind. ZECA (Zeno Engaging Children with

ICT4AWE 2021 - 7th International Conference on Information and Communication Technologies for Ageing Well and e-Health

Autism) has a child-like appearance with 34 degrees

of freedom: 4 being located in each arm, 6 in each leg,

11 in the head, and 1 in the waist. The robot is capable

of expressing facial expressions thanks to the servo

motors mounted on its face and a special material,

Frubber, which has a similar look and feel to human

skin.

Figure 1: Proposed system setup. Starting from the left: the

developed OPT (PlayBrick), central processing unit, and

the humanoid robot ZECA.

The Robokind software performs animation and

motion control functions and it includes an

Application Programming Interface (API) for rapid

integration of other components and shared control.

Regarding the OPT, the design approach consisted

in developing a device that can offer a tangible

experience, adapt itself to different games scenarios

and provide immediate feedback. It was designed with

the purpose of being used in different activities and

contexts by physically (re)programming it or by adding

new components such as sensors. This may offer an

exceptional opportunity to measure behaviours since

children with ASD are less willing to use wearable

devices (Bekele, Crittendon, Swanson, Sarkar, &

Warren, 2014), which may be a challenge when trying

to extract additional behaviour information. The

addition of feedback from the OPT to the user is a key

feature that allows guiding the child through the play

activity and to enjoy the experience in all its fullness.

Since the users expect to see and feel the results of their

actions, the immediate feedback feature is a very

important factor specially when designing devices for

children with impairments. Furthermore, learning via

reinforcement can be one of the most effective

approaches to reinforcing desired behaviours,

particularly with children with ASD, allowing the

formation of an association between a suggestion or

action and a reinforcement with some intrinsic

motivational value (Schuetze, Rohr, Dewey,

McCrimmon, & Bray, 2017). Since the experience

should be configurable, the type of feedback should

also be adaptable according to the child preferences.

For example, some types of feedback may be

uncomfortable for some children with ASD, such as

the sound feedback that may be unenjoyable since

some individuals are oversensitive to environmental

stimuli (Schoen, Paul, & Chawarska, 2011).

Following these ideas, the developed OPT,

PlayBrick (Fig. 2), was designed to provide a tangible

and adaptive experience, being easy and intuitive to

manipulate through natural gestures (such as touch,

tilt, rotation), with different sources of immediate

feedback (haptic and visual). The PlayBrick has a 5.0-

inch touch screen, an Inertial Measurement Unit

(IMU), a haptic driver with a Linear Resonant

Actuator (LRA), and a LED RGB strip. It has built-in

Bluetooth and Wi-Fi communication.

Figure 2: General view of the PlayBrick and its main

components.

2.2 Designing the End User Activity

The developed activities are focused on emotion

recognition tasks. The first task, denominated

recognize, consists in the robot displaying randomly

one of the five basic facial expressions (happiness,

sadness, anger, surprised, and afraid) and its

associated gestures (body posture), Fig. 3. Then, the

child is prompted to identify the emotion associated

with the facial expression.

Figure 3: The different facial expressions displayed by the

robot ZECA: a) anger, b) fear, c) happiness, d) surprise, and

e) sadness (Soares et al., 2019).

The second task is the storytelling game, in which

randomly selected stories among 15 social stories are

told by the robot, and the participant has to identify

the emotion of the main actor, i.e. the robot (Fig. 4).

The identification of emotions displayed by other

people is essential (Clark, Winkielman, & McIntosh,

2008), being fundamental for successful social

interactions. As most children with ASD have

Hybrid Approach to Promote Social Interaction with Children with Autism Spectrum Disorder

alterations in the central auditory processing, vision

is one of the strongest skills of individuals with ASD

(Caldeira da Silva et al., 2012), meaning that using a

visual support can be fundamental in helping children

with ASD understand requests and tasks. Therefore,

each social story has its associate visual

representation of the story scenario. As the robot

starts telling the story, an image is simultaneously

shown representing the social context of the story as

a visual cue (Fig. 4). Then, the child is prompted to

answer how the robot felt in that story scenario.

In both game scenarios, the child selects the

answer by tilting back and forward the PlayBrick,

scrolling through the facial expressions (common

emoji) displayed by the OPT and touching the image.

Figure 4: Sample of images for the storytelling game

scenario. The scenario A represents the sad emotion, with

the following story: “I like to play when I’m at home. Today

I took my ball and played with the ball in the living room. I

kicked strongly the ball and broke a window. My mother me

scolded me and I cried.” The visual cue B represents the

fear emotion, with the following story: “I go with my mother

shopping. I like to choose the yogurts that I eat. Today, at

the exit of the supermarket, a very large dog began barking

very loud. I was shaking.”

The type of input is also configurable, meaning

that if a child has difficulties in manipulating (tilting)

the device, the interface changes by showing two

arrows that also allow the participant to search and

select the possible answer (Fig. 5). In parallel, when

the answer is selected, a positive or negative

reinforcement is prompted by ZECA and the

PlayBrick. The type of reinforcement is configurable

on both the robot and the OPT according to the child

preferences.

Figure 5: Prompted interface to the child on the OPT screen

to select the answer. The arrows are visible in case the child

has difficulties in manipulating the PlayBrick.

2.3 Procedures and Participants

Tests following the experimental design in Fig. 6

were conducted in a school environment in a triadic

setup, i.e., child-robot-researcher. The goal of the

pilot study was to detect the systems constraints and

verify if the system can implement a procedure that

makes the children able to interact in a comfortable

and natural way during an intervention session.

In the experimental setup used in the school, the

child sits in front of the robot, which is positioned at

approximately 85 cm from the child’s line of sight.

Two cameras are placed behind the robot in order to

video record the sessions. One camera (A) records

only the child and camera B records the overall

session. The researcher is seated next to the child in

order to assist the child during the task. This layout is

proposed in order to establish a basis of comparison

between the participants along the sessions, since the

experiments are conducted in an unconstrained (in

this case a school) but familiar setting for the child.

All procedures involving the children with ASD

during the study were conducted in accordance with

the following ethical concerns: the research work was

approved by the ethical committee of the university

of Minho, a collaboration protocol was firmed

between the university and the school, and informed

consents were signed by the parents/tutors of the

children that participated in the studies.

A sample of 4 children previously diagnosed with

ASD aged 6 to 10 years (M= 8.75; SD=0.96) was

selected to participate in this study. From here on, the

children are identified as A, B, C, and D. Although

ASD is more prevalent in boys (Christensen et al.,

ICT4AWE 2021 - 7th International Conference on Information and Communication Technologies for Ageing Well and e-Health

2016), the selected sample has as many boys (2) as

girls (2). All children that participated in the study are

verbal but their attempts to initiating interactions and

make friends are odd and typically unsuccessful –

which corresponds to level 1 of severity levels of

DSM5 in 2013. The participants are high functioning,

according to their diagnosis.

Figure 6: The experimental design used during the

experiments in a triadic configuration: child, robot,

researcher.

According to the therapists, the type of reinforcement

used in the study was the same for three children

(robot: verbal + movement + sound; PlayBrick: visual

+ haptic) and different from one child (robot: verbal +

movement; PlayBrick: visual + haptic).

The study was carried out during four sessions

spaced one week between the second and the third

session. In the first session the children played the

recognize game scenario. The storytelling activity

was conducted in the 3 remaining sessions (Fig. 6).

2.4 Analysis

The videos of the experiments of the 4 sessions were

coded by one observer specialized in behavioural

psychology. To assess the children’s engagement

during the activity, the frequency of children’s gaze

towards to the robot and to the PlayBrick as well as

the duration of such events were registered. The

number of times that the children needed help and the

number of wrong usages of the PlayBrick were

counted. Additionally, the number of correct,

incorrect, and unanswered answers as well as the

number of total robot prompts during the sessions

were quantified. The children’s mean response time

to the robot’s prompts were also registered. At the end

of each session, the robot asked the participant if

he/she wanted to play more. This was also quantified.

The non-parametric Wilcoxon signed-rank test

(alternative to parametric paired t-test) was used to

compare the children’s attention during the

storytelling game scenario. This test will be reported

by using the Z statistic.

3 RESULTS

A set of experiments were carried out in a school

setting involving four children with ASD (children A,

B, C, and D). Both activities (recognize and

storytelling) were played.

Regarding the children’s attention it was found

that, in general, children gazed more towards to the

robot in the first session (recognize game scenario),

Fig. 7. In the other sessions (storytelling) the children

tended to gaze more at the OPT device.

Furthermore, the children’s mean duration per

gaze, in general, was higher towards the PlayBrick

than towards the robot (Fig. 8). In particular, the

children’s mean duration per gaze in the storytelling

game scenario was significantly less for the gazes

directed at the robot (M=5.28; SD=1.76) than those

directed at the PlayBrick (M=12.58; SD=2.97), Z=-

3.059, p<0.001.

Figure 9 shows the number of right and wrong

answers, as well as the number of no answered

prompts. In general, it is possible to notice a positive

evolution, regarding the number of successful

answers along the sessions. Moreover, the

participants answered all of the robot’s prompts.

In general, the children’s mean response time to

the robot prompts increased between the first session

(where the recognize activity occurred) and the

second session (where the storytelling game scenario

was played), Fig. 10-A. Regarding sessions 3 and 4,

the children’s mean response time remain, in general,

unchanged.

Additionally, the number of times that a child

needed help (Fig. 10-B) on how to manipulate the

PlayBrick decreased over the sessions.

During the study only one child manipulated

wrongly the PlayBrick, two times, being assisted by the

researcher on how to correctly manipulate de device.

Table 1 shows the sessions where each participant

answered positively (‘✓’) to the robot asking if he/she

wants to play more.

Table 1: The sessions where the children wanted to

continue the activity (‘

✓ ’).

Child Session

1 2 3 4

✓ ✓ ✓ -

✓ - - -

- ✓ - -

✓ ✓ ✓

Hybrid Approach to Promote Social Interaction with Children with Autism Spectrum Disorder

Figure 7: Percentage and number of gazes towards the robot and the PlayBrick for the 4 children (A, B, C, and D) during the

sessions for the game scenarios recognize (session 1) and storytelling (sessions 2, 3, and 4). It is possible to perceive that,

overall, the children gaze more towards to the robot in the first session. In the remaining sessions, children tended to look

more at the PlayBrick.

Figure 8: Mean gaze time towards the robot and the PlayBrick for the 4 children (A, B, C, and D) during the sessions for the

game scenarios recognize (session 1) and storytelling (sessions 2, 3, and 4).

Figure 9: Children’s answers to the robot prompts during the four sessions.

20%

40%

60%

80%

100%

1234 1234 1234 1234

ABCD

Percentage of Gazes

Look at the robot(B1) Look at the OPT(B2)

1234 1234 1234 1234

AB CD

Mean gaze time (s)

Look to the robot(B1) Look to the OPT(B2)

1234 1234 1234 1234

ABCD

Number of answers

#RightAnswers #WrongAnswers

ICT4AWE 2021 - 7th International Conference on Information and Communication Technologies for Ageing Well and e-Health

Figure 10: A – Children’s mean response time in seconds (with confidence interval) to the robot prompts during the four

sessions. B – Number of times that each child need help to interact with the PlayBrick over the four sessions.

4 DISCUSSION

Concerning children’s attention in the activity, there

were more gazes towards the robot during the first

session compared to the other sessions (Fig. 7), since

during this activity the child had to look at the robot

face in order to identify the robot facial expression.

However, the mean gaze time towards to the robot, in

general, was lower when compared to the mean gaze

time towards the PlayBrick, Fig. 8. The mean gaze

time towards the OPT was significantly higher (p-

value lower than 0.05 significance level) during the

sessions where the storytelling game was conducted,

indicating that the children relied on the storytelling

visual cues displayed by the PlayBrick. This supports

that the use of visual cues may be paramount for

children with ASD to understand the tasks. In

addition, the children answered all prompts and the

number of right answers were superior to the number

of wrong answers (Fig. 9), further supporting that the

children understood the games and successfully

interacted with the PlayBrick. Moreover, the children

rapidly adapted and learned how to interact with the

OPT and the robot, since the number of times that

each child needed help during the activities decreased

over the sessions, Fig. 10-B.

The mean response time to the robot prompts

during the storytelling activity was higher when

compared to the recognize activity (Fig. 10-A), which

is expected since during this activity the child has to

identify the state of mind of the main character of the

story.

In general, all children wanted to continue the

activities (Table 1), meaning that they enjoyed and

were successfully engaged in the activities.

Additionally, it is worth mentioning that none of the

children that participated in this study ever abandoned

an activity.

5 CONCLUSIONS

The comprehension of social emotional cues is

important for a successful human communication.

However, individuals with ASD present impairments

in social communication. New forms of intervention

have been explored and conducted in the last years by

employing the use of technological devices trying to

mitigate the emotion recognition impairments that

children with ASD present.

Following this idea, the present work shows the

developments of a hybrid approach in human-robot

interaction with children with ASD in emotion

recognition tasks. This hybrid approach includes the

OPT device PlayBrick, used as an add-on to the

humanoid robot intervention.

By analysing the results, it is possible to conclude

that the children understood the mechanics of the

games and successfully interacted with PlayBrick.

There was a significant difference in the mean gaze

time towards the OPT, particularly in the storytelling

scenario, suggesting that the children used/relayed on

the PlayBrick during the activities. Moreover, in

general, the children were keen to participate in the

activities since they wanted to continue on playing.

Additionally, they were also attentive to the

PlayBrick feedback, lights, haptic as well as the

images for correct and incorrect answers displayed on

the screen.

The future work includes further improvements of

this hybrid approach. A study will be conducted

involving a larger sample of children with ASD to

understand if and how the presented method may be

used as a valuable tool to promote social interaction

with children with ASD.

1234

Mean response time

(s)

Session

A B C D

-1

1234

Number of help times

Session

A B C D

Hybrid Approach to Promote Social Interaction with Children with Autism Spectrum Disorder

ACKNOWLEDGEMENTS

This work has been supported by FCT – Fundação

para a Ciência e Tecnologia within the R&D Units

Project Scope: UIDB/00319/2020. Vinicius Silva

thanks FCT for the PhD scholarship SFRH/BD/

SFRH/BD/133314/2017. The authors thank the

teachers and students of the Elementary School of

Gualtar (EB1/JI Gualtar) in Braga for their

participation in the study.

REFERENCES

American Psychiatric Association, 2013. Diagnostic and

Statistical Manual of Mental Disorders, 5th Edition

(DSM-5), Diagnostic and Statistical Manual of Mental

Disorders 4th edition TR., p. 280. doi:

10.1176/appi.books.9780890425596.744053.

Barajas, A. O., Al Osman, H. and Shirmohammadi, S.,

2017. A Serious Game for children with Autism

Spectrum Disorder as a tool for play therapy, in 2017

IEEE 5th International Conference on Serious Games

and Applications for Health, SeGAH 2017. doi:

10.1109/SeGAH.2017.7939266.

Bekele, E. et al., 2014. Pilot clinical application of an

adaptive robotic system for young children with autism,

Autism : the international journal of research and

practice, 18(5), pp. 598–608. doi:

10.1177/1362361313479454.

Boucenna, S. et al., 2014. Interactive Technologies for

Autistic Children: A Review, Cognitive Computation,

6(4), pp. 722–740. doi: 10.1007/s12559-014-9276-x.

Breazeal, C., 2000. Sociable machines: Expressive social

exchange between humans and robots, Expressive

Social Exchange Between Humans and Robots.

Available at:

http://groups.csail.mit.edu/lbr/mars/pubs/phd.pdf.

Caldeira da Silva, P. et al., 2012. Programa clínico para o

tratamento das perturbações da relação e da

comunicação, baseado no Modelo D.I.R., Análise

Psicológica. doi: 10.14417/ap.116.

Christensen, D. L. et al., 2016. Prevalence and

Characteristics of Autism Spectrum Disorder Among

Children Aged 8 Years — Autism and Developmental

Disabilities Monitoring Network, 11 Sites, United

States, 2012, MMWR. Surveillance Summaries, 65(3),

pp. 1–23. doi: 10.15585/mmwr.ss6503a1.

Clark, T. F., Winkielman, P. and McIntosh, D. N., 2008.

Autism and the Extraction of Emotion From Briefly

Presented Facial Expressions: Stumbling at the First

Step of Empathy, Emotion. doi: 10.1037/a0014124.

Dautenhahn, K. and Werry, I., 2004. Towards interactive

robots in autism therapy: Background, motivation and

challenges, Pragmatics & Cognition, 12(1), pp. 1–35.

doi: 10.1075/pc.12.1.03dau.

Ferrari, E., Robins, B. and Dautenhahn, K., 2009.

Therapeutic and educational objectives in robot assisted

play for children with autism, IEEE International

Workshop on Robot and Human Interactive

Communication (Ro-Man), pp. 108–114. doi:

10.1109/ROMAN.2009.5326251.

Fujimoto, I. et al., 2011. Mimicking and evaluating human

motion to improve the imitation skill of children with

autism through a robot, International Journal of Social

Robotics, 3(4), pp. 349–357. doi: 10.1007/s12369-011-

0116-9.

Gillesen, J. C. C. et al., 2011. From training to robot

behavior: Towards custom scenarios for robotics in

training programs for ASD, in IEEE International

Conference on Rehabilitation Robotics. doi:

10.1109/ICORR.2011.5975381.

Kim, E. et al., 2012. Bridging the Research Gap: Making

HRI Useful to Individuals with Autism, Journal of

Human-Robot Interaction, pp. 26–54. doi:

10.5898/JHRI.1.1.Kim.

Lund, H. H., Dam Pedersen, M. and Beck, R., 2009.

Modular robotic tiles: Experiments for children with

autism, Artificial Life and Robotics, 13(2), pp. 394–

400. doi: 10.1007/s10015-008-0623-4.

Lund, H. H., Klitbo, T. and Jessen, C., 2005. Playware

technology for physically activating play, Artificial

Life and Robotics, 9(4), pp. 165–174. doi:

10.1007/s10015-005-0350-z.

Lund, H. H. and Marti, P., 2009. Designing modular robotic

playware, Proceedings - IEEE International Workshop

on Robot and Human Interactive Communication, pp.

115–121. doi: 10.1109/ROMAN.2009.5326286.

Mazzei, D. et al., 2011. Development and evaluation of a

social robot platform for therapy in autism, Engineering

in Medicine, Proceedings of the Annual International

Conference of the IEEE, 2011, pp. 4515–8. doi:

10.1109/IEMBS.2011.6091119.

Michaud, F. et al., 2005. Autonomous spherical mobile

robot for child-development studies, IEEE Transactions

on Systems, Man, and Cybernetics - Part A: Systems

and Humans, 35(4), pp. 471–480. doi:

10.1109/TSMCA.2005.850596.

Pennisi, P. et al., 2016. Autism and social robotics: A

systematic review, Autism Research, pp. 165–183. doi:

10.1002/aur.1527.

Schoen, E., Paul, R. and Chawarska, K., 2011. Phonology

and vocal behavior in toddlers with autism spectrum

disorders, Autism Research. doi: 10.1002/aur.183.

Schuetze, M. et al., 2017. Reinforcement learning in autism

spectrum disorder, Frontiers in Psychology. doi:

10.3389/fpsyg.2017.02035.

Silva, V. et al., 2018. Building a hybrid approach for a game

scenario using a tangible interface in human robot

interaction, in Lecture Notes in Computer Science

(including subseries Lecture Notes in Artificial

Intelligence and Lecture Notes in Bioinformatics). doi:

10.1007/978-3-030-02762-9_25.

Soares, F. O. et al., 2019. Socio-emotional development in

high functioning children with Autism Spectrum

Disorders using a humanoid robot, Interaction Studies

Social Behaviour and Communication in Biological

ICT4AWE 2021 - 7th International Conference on Information and Communication Technologies for Ageing Well and e-Health

and Artificial Systems, 20(2), pp. 205–233. doi:

10.1075/is.15003.cos.

Tapus, A. et al., 2012. Children with autism social

engagement in interaction with Nao, an imitative

robot’, Interaction Studies, 13(3), pp. 315–347.

Tapus, A., Member, S. and Scassellati, B., 2007. The Grand

Challenges in Socially Assistive Robotics, IEEE

Robotics and Automation Magazine, 14, pp. 1–7. doi:

10.1109/MRA.2010.940150.

Hybrid Approach to Promote Social Interaction with Children with Autism Spectrum Disorder