A Review of the NAO Robot in Treatments for Children with Autism Spectrum Disorder
Introduction
The Centers for Disease Control (CDC) estimates that one in thirty-six children have Autism Spectrum Disorder (ASD) and autism prevalence has increased since 2014 [1, 2]. Children with ASD have poor communication skills, social skills, and stimuli processing [3, 4, 5]. Though ASD has a strong genetic basis, doctors can only reliably diagnose children based on their developmental and behavioral history [4, 5, 6]. Thus, the socio-communicative deficits that children with ASD face are treated at the behavioral level. Nowadays, popular interventions for children with ASD target behavioral, educational, and socio-relational improvement [7]. To reduce the workload of human therapists, robots have been used in recent years to record behaviors, engage children, and adapt levels of treatment [8].
The most studied robot in ASD interventions is NAO. NAO is a plastic-bodied, 58-centimeter-tall humanoid robot with 25 different directions of programmable human-like movements [9, 10, 11, 12, 13]. NAO has cameras, microphones, and sensors to measure patient interactions and LED lights and loud-speakers to respond to patient interactions [10, 14]. NAO is also equipped with a text-to-speech engine, and sound detection to recognize a patient's speech and verbally communicate with a patient in real time [9, 15]. Overall, the features of NAO, from its humanoid shape to its programmability, non-verbal and verbal methods of communication, and recording capabilities make it the most popular choice of robot in recent robot-related ASD therapy. This review analyzes studies done on the interaction of NAO with human therapists, parents, and children to improve joint attention, motor skills, and prosocial behaviors of children with ASD.
Standalone Robot Therapy
Joint Attention
Studies on standalone robot therapy investigate the effectiveness of robot-mediated therapies on the joint attention of children with ASD. Joint attention is the act of two people paying attention to the same stimulus intentionally, for social reasons. In seven 15-minute sessions, Rakhymbayeva et al. (2021) computer-controlled and programmed NAO with human-like movements and audio cues for the child to respond to. The study participants consisted of one girl and ten boys between four and eleven years old, all diagnosed with ASD [13]. This study investigated the effectiveness of long-term NAO therapy by analyzing the children's engagement in various social activities with NAO throughout their sessions [13]. Although there was no statistically significant change in engagement duration over the seven sessions, there was significantly higher engagement in familiar tasks compared to tasks the children had not yet completed [13].
In an additional analysis, Alnajjar F et al. (2020) also computer-controlled NAO to point and audibly direct the child's joint attention to another object. They achieved this by utilizing a mobile device displaying facial expressions mounted on NAO to detect the child's attention on the robot [16]. The study participants consisted of eleven boys under the age of sixteen with a mean age of nine years, all diagnosed with ASD [16]. Over the course of four 5-minute sessions, the children's attention to NAO generally increased over time with more sessions [16]. Any fluctuations in attention were attributed to external factors, such as the child skipping medication or missing energy-draining physical therapy exercises meant to calm the child prior to the session [16].
These studies indicate the need for children to become familiar with NAO through longer-term interventions. These interventions aim to improve NAO-based ASD therapies that target joint attention. Additionally, the studies highlight that NAO-based joint attention therapies depend on parents providing their children with consistent transportation to these sessions, as well as behavioral medication and physical therapies to optimize the time spent with NAO.
Motor Skills
Standalone robot therapy also targeted the motor skills of children with ASD. Feng H et al. (2022) studied the effects of music-therapy by programming NAO to autonomously play the xylophone over six sessions. The participants included one girl and eight boys with a mean age of twelve years, diagnosed with high functioning ASD [11]. Most of these participants completed the motor control task of copying NAO's xylophone sequence with 70% accuracy [11]. Over half of the participants were more focused on playing their favorite songs even though the difficulty of the sequence increased [11].
In a subsequent study, Conti D et al. (2021) used a computer-controlled NAO over fourteen sessions to conduct a motor imitation intervention. The participants comprised six boys between five and ten years old with a mean age of nine years, diagnosed with both ASD and intellectual disability [8]. Four out of the six participants learned to perform and improve their motor imitation skills [8]. The two children unsuccessful with the motor tasks were diagnosed with severe intellectual disability and expressed difficulty understanding the stimuli given in the intervention [8]. The children who successfully completed the robot therapy were able to engage in new imitation tasks with human therapists, and they maintained their imitation skills for three months [8].
Both studies show that there is variation in the effectiveness of NAO therapies for improving motor skills in children with ASD. Variations in these children's abilities should be accounted for in future studies with NAO to make personalized interventions that cultivate attainable and sustainable therapeutic goals for children. Especially for children successful in the interventions, these studies support NAO being a motivational tool to attain and improve communication and motor skills in a long-term therapy format.
Standalone Robot Therapy with Parent Involvement
Researchers in the following studies invited parents to participate in their children's standalone robot therapy. Although the results of the studies did not find the influence of parent involvement significant in their children's performance in the therapy sessions, researchers did make findings that inform future studies of parent involvement in NAO therapy. Over five virtual sessions, Javed H et al. (2022) programmed NAO to autonomously dance based on the child's pose and facial expressions. The participants consisted of one ten-year-old boy, one eleven-year-old boy, and one seven-year-old girl, all diagnosed with ASD [17]. This study examined how well the NAO robot promoted physical activity in the children by comparing their engagement in three different types of sessions: NAO as a leader for the child to follow, NAO as a follower by mimicking the child's movements, and NAO acting as both the leader and follower [17]. The researchers found no significant difference in engagement between the sessions in which NAO acted both as a leader and follower and the sessions in which NAO acted as either a leader or a follower [17]. However, the researchers did find that as a result of allowing the parents to join the therapy sessions, creative parents encouraged their child to practice verbal language skills in addition to practicing motor skills through dancing [17].
In a different study, Amirova et al. (2022) computer-controlled NAO to implement behavioral therapy over the course of two to nine 15-minute sessions. The participants consisted of two girls and fourteen boys between five and twelve years old, with a mean age of seven [15]. All diagnosed with ASD and eight of the participants also diagnosed with ADHD [15]. The NAO-administered behavioral therapy consisted of dancing, singing, imitation, and social interaction activities [15]. The study found that there was no significant influence of parent's presence on children's engagement with NAO during the behavioral activities [15]. The researchers did find that parents comforted children distressed with NAO, but parental overinvolvement weakened the effectiveness of the NAO intervention for the verbal children [15].
Overall, parents engaged in their children's therapy sessions seemed to be beneficial for improving skills not directly targeted by the NAO intervention, as well as providing comfort with an unfamiliar NAO. Future NAO interventions should consider keeping but limiting parent involvement based on a child's communication abilities to allow these children to practice autonomy by expressing their own needs independently.
Robot vs. Human Therapy
Evidence Favoring NAO Intervention
Studies indicate that robot therapy is more effective than human-only interventions targeting the socio-communication skills of children with ASD. In biweekly sessions over three months, Holeva V et al. (2022) used exercises that targeted turn taking, conversation, emotional self-regulation, and problem-solving skills. The participants consisted of forty-four children with a mean age of nine, all diagnosed with ASD [14]. The robot group had nineteen boys and three girls, and the control group had sixteen boys and six girls [14]. The robot group consisted of the child, the programmed NAO, and the therapist while the control group only included the child and the therapist [14]. Researchers found statistically significant differences in favor of the NAO group over the control group for improvement of eye contact, speech interaction, and gesture interaction [14].
In a second study, van den Berk-Smeekens I et al. (2021) used a three-step intervention: introducing a stimulus, targeting a behavior, and reinforcing the target behavior, to improve children's communication skills over the course of twenty 45-minute weekly sessions. The participants consisted of fifty children with a mean age of six, all diagnosed with ASD [18]. The robot-intervention group had twenty boys and five girls, and the control group had twenty-two boys and three girls [18]. The robot-intervention group included 15 minutes with a computer-controlled NAO and parent-child interaction while the control group involved a human therapist and parent-child interaction [18]. Researchers found significant improvements in social-communication skills in the followup assessment of the participants in the robot-assisted group compared to the control group [18].
In another study, Marino F et al. (2019) used goal-oriented tasks to improve emotional regulation over the course of ten 90-minute sessions. The participants consisted of fourteen children with a mean age of six, all diagnosed with ASD [12]. The robot-assisted group and the control group both had six boys and one girl each [12]. NAO, programmed for emotional and social prompts via body movements and verbal scripts, acted as a partially-controlled co-therapist in the robot-assisted group [12]. The control group were given the same emotional regulation tasks as the robot-assisted group but without the use of NAO [12]. Researchers found significant differences in socio-emotional understanding skills in favor of the NAO intervention over the human intervention [12].
In all, these studies favor the effectiveness of long-term robot intervention over human-mediated therapy for improving communication skills of children with ASD that encompass eye contact, gestures, and emotional understanding. These studies also support researchers spending time on programming NAO to enhance NAO socio-emotional interventions, which have been proven more effective than human intervention alone.
Evidence Favoring Human Intervention
Although previous studies favored the effectiveness of robots over human intervention in targeting socio-communication skills, these studies found that human intervention is more effective than robot intervention for targeting attention. Baraka K et al. (2022) computer-controlled NAO for a joint attention task and a response to name calling task as measures for attention during 15-minute sessions. The participants consisted of seven boys and four girls aged between two and seven years, all diagnosed with ASD [9]. Researchers found that the children generally needed more prompts from the robot than the human therapist to complete either joint attention or the name calling task successfully [9].
In an alternative study, David DO et al. (2018) used prompting and reinforcement to teach children with ASD how to respond to a partner's stimulus with appropriate behaviors. The participants consisted of four boys and one girl aged between three and five years diagnosed with ASD [10]. This intervention involved twenty sessions: eight sessions with a computer-controlled NAO as the therapist, eight sessions with a human therapist, and four sessions with whichever therapist was deemed more statistically effective [10]. Effectiveness was statistically determined by comparing the child's improvement in the human therapist intervention against the NAO intervention [10]. Joint attention was scored via points given to correct head orientation, pointing, and relevant verbalization and taken away for delay of the child's response [10]. Both the robot and control group included the child, the therapist (human or NAO), and a human mediator [10]. Researchers found that more cues, such as using head turning, pointing, and vocal instruction in combination, lead to increased performance in joint attention for three of the five participants [10]. The two other participants showed improvement in response to the human intervention, but not the robot intervention [10].
Overall, these studies indicate that robot therapy is not yet optimized for increasing the attention span of children with ASD. This suggests that human therapists work in tandem with NAO for the attention exercises or NAO not be used at all in attention-improvement activities.
Conclusion
The purpose of this review was to compile recent studies of the NAO robot in ASD therapies for children. From the reviewed literature, the uses of NAO cover a wide range of different therapies and human interactions. These therapies aim to enhance emotional, motor, imitation, and social skills. Various studies have demonstrated NAO used in different therapy dynamics, including NAO working solely with the child, NAO and the parent both interacting with the child, or NAO engaging the child alongside a human therapist. Given the small sample sizes and variations in intervention duration across studies, generalizing the results becomes challenging, especially with conflicting reports of how robot-therapy compares to human administered therapy. NAO was generally more effective at administering replicative therapies that target motor and imitation skills, while human therapists excelled in attention span exercises. Future studies will likely move towards long-term NAO robots and individualizing therapies for motor and communication skills. Improvements in NAO therapy will likely motivate parents and therapists to administer attention regulation treatments in the form of medication and vigorous physical therapy to optimize the child's development during NAO treatments.
About the Author: Stephanie Vu
Stephanie Vu is a third year Neurobiology, Physiology, and Behavior major, pursuing a minor in Human Development. Reflecting on the experience of growing up with her two brothers with severe autism, she was inspired to research novel autism spectrum disorder (ASD) interventions. Upon reading many research articles and narrowing down her topic, Stephanie found that many researchers today focus on the NAO robot to implement interventions that supplement human therapy techniques for children with ASD. Though the NAO robot is a niche treatment for an increasingly prevalent disorder, Stephanie hopes to bring more awareness to ASD as a whole, as well as inspire more innovation for its treatments.
Author's Note
Wrote for Lit Review project for UWP 102B, Fall Quarter 2023; recommended from professor to submit to Aggie Transcript. Interested in treatments/interventions for individuals with autism spectrum disorder because I have two brothers with severe ASD conditions. By narrowing down my ASD topic, I found that robot interventions for individuals with ASD is a newly researched topic. Searching articles with the NAO robot therapy for children with ASD ended up being the final topic since there was a good balance between number of articles published in the last five years and specificity. Main takeaways:
ASD prevalence increased in children in recent years
ASD can currently only be diagnosed and treated at the behavioral level by age 2
NAO robot addition to current behavioral treatments prove to be most effective for motor and communication skill and in long-term format - current research has tested the effectiveness of the NAO robot in interventions that target the attention span, communication skills, and motor skills of children with ASD
References
Centers for Disease Control and Prevention. 2023. Data & statistics on autism spectrum disorder. Centers for Disease Control and Prevention. https://www.cdc.gov/ncbddd/autism/data.html.
Chiarotti F, Venerosi A. 2020. Epidemiology of Autism Spectrum Disorders: A Review of Worldwide Prevalence Estimates Since 2014. Brain Sciences. 10(5). doi:https://doi.org/10.3390/brainsci10050274. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7288022/.
Centers for Disease Control and Prevention. 2022. Autism Spectrum Disorder Diagnostic Criteria. Centers for Disease Control and Prevention. https://www.cdc.gov/ncbddd/autism/hcp-dsm.html.
Centers for Disease Control and Prevention. 2022. What is autism spectrum disorder? Centers for Disease Control and Prevention. https://www.cdc.gov/ncbddd/autism/facts.html.
Sauer AK, Stanton JE, Hans S, Grabrucker AM. 2021. Autism Spectrum Disorders: Etiology and Pathology. Autism Spectrum Disorders.:1–16. doi:https://doi.org/10.36255/exonpublications.autismspectrumdisorders.2021.etiology. https://www.ncbi.nlm.nih.gov/books/NBK573613/.
Centers for Disease Control and Prevention. 2022. Screening and Diagnosis of Autism Spectrum Disorder. Centers for Disease Control and Prevention. https://www.cdc.gov/ncbddd/autism/screening.html.
Centers for Disease Control and Prevention. 2022. Treatment and intervention services for autism spectrum disorder. Centers for Disease Control and Prevention. https://www.cdc.gov/ncbddd/autism/treatment.html.
Conti D, Trubia G, Buono S, Di Nuovo S, Di Nuovo A. 2021. An empirical study on integrating a small humanoid robot to support the therapy of children with Autism Spectrum Disorder and Intellectual Disability. Interaction Studies. 22(2):177–211. doi:https://doi.org/10.1075/is.21011.con.
Baraka K, Couto M, Melo FS, Paiva A, Veloso M. 2022. “Sequencing Matters”: Investigating Suitable Action Sequences in Robot-Assisted Autism Therapy. Frontiers in Robotics and AI. 9. doi:https://doi.org/10.3389/frobt.2022.784249.
David DO, Costescu CA, Matu S, Szentagotai A, Dobrean A. 2018. Developing Joint Attention for Children with Autism in Robot-Enhanced Therapy. International Journal of Social Robotics. 10(5):595–605. doi:https://doi.org/10.1007/s12369-017-0457-0.
Feng H, Mahoor MH, Dino F. 2022. A Music-Therapy Robotic Platform for Children With Autism: A Pilot Study. Frontiers in Robotics and AI. 9. doi:https://doi.org/10.3389/frobt.2022.855819.
Marino F, Chilà P, Sfrazzetto ST, Carrozza C, Crimi I, Failla C, Busà M, Bernava G, Tartarisco G, Vagni D, et al. 2019. Outcomes of a Robot-Assisted Social-Emotional Understanding Intervention for Young Children with Autism Spectrum Disorders. Journal of Autism and Developmental Disorders. 50(6):1973–1987. doi:https://doi.org/10.1007/s10803-019-03953-x.
Rakhymbayeva N, Amirova A, Sandygulova A. 2021. A Long-Term Engagement with a Social Robot for Autism Therapy. Frontiers in Robotics and AI. 8. doi:https://doi.org/10.3389/frobt.2021.669972.
Holeva V, Nikopoulou VA, Lytridis C, Bazinas C, Kechayas P, Sidiropoulos G, Papadopoulou M, Kerasidou MD, Karatsioras C, Geronikola N, et al. 2022. Effectiveness of a Robot-Assisted Psychological Intervention for Children with Autism Spectrum Disorder. Journal of Autism and Developmental Disorders.:1. doi:https://doi.org/10.1007/s10803-022-05796-5. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC9638397/.
Amirova A, Rakhymbayeva N, Zhanatkyzy A, Telisheva Z, Sandygulova A. 2022. Effects of Parental Involvement in Robot-Assisted Autism Therapy. Journal of Autism and Developmental Disorders. doi:https://doi.org/10.1007/s10803-022-05429-x.
Alnajjar F, Cappuccio M, Renawi A, Mubin O, Loo CK. 2020. Personalized Robot Interventions for Autistic Children: An Automated Methodology for Attention Assessment. International Journal of Social Robotics. doi:https://doi.org/10.1007/s12369-020-00639-8.
Javed H, Park CH. 2022. Promoting Social Engagement With a Multi-Role Dancing Robot for In-Home Autism Care. Frontiers in Robotics and AI. 9. doi:https://doi.org/10.3389/frobt.2022.880691.
van den Berk-Smeekens I, de Korte MWP, van Dongen-Boomsma M, Oosterling IJ, den Boer JC, Barakova EI, Lourens T, Glennon JC, Staal WG, Buitelaar JK. 2021. Pivotal Response Treatment with and without robot-assistance for children with autism: a randomized controlled trial. European Child & Adolescent Psychiatry. doi:https://doi.org/10.1007/s00787-021-01804-8.