Social Robots as Assistants: New Approaches to Supporting Children with Autism

Early, individually tailored therapeutic approaches are crucial for supporting children with Autism Spectrum Disorder (ASD). The WHO emphasizes that disability does not arise solely from personal impairments but is exacerbated by societal barriers. When these barriers - such as those reduced by technical aids - are removed, participation can be significantly improved.

ASD affects approximately 1 in 100 to 160 children worldwide. The rise in diagnosis rates is mainly due to improved diagnostic methods and increased societal awareness. Children with ASD often exhibit atypical social interactions and are sensitive to sensory stimuli. This is where social robots come in: they create structured, predictable routines that make it easier for children with ASD to engage in social interactions. In therapy, they specifically promote social, emotional, and communicative skills such as eye contact, imitation, or emotion recognition. Robots are intended to support professionals, but never to replace them. It is important that they are easy to use and flexible in application.

Thanks to advances in artificial intelligence, research interest in robot-assisted therapy is growing. Humanoid, animal-like, or toy-like robots are used, such as NAO, KASPAR, Keepon, or Cozmo. Their use ranges from structured individual therapy to playful group settings. An inclusive research approach, involving professionals, families, and individuals with ASD working together to develop interventions, has proven particularly promising.

In the study by Ghiglino et al. (2021), 24 children with ASD, averaging 5.8 years old, received robot-assisted training with the toy robot Cozmo alongside classical therapy-in varying sequences. The combination of both approaches led to a significant improvement in initiating social interactions. A clear effect was also observed in behavioral requests. The study illustrates that robot-assisted interventions can be a meaningful complement to established therapy methods.

Huijnen et al. (2017) investigated the requirements for robot-assisted interventions with the robot KASPAR through focus groups and co-creation sessions involving professionals, family members, and individuals with ASD. The results were incorporated into a practical intervention concept. Particularly important were: child-appropriate design, ease of use, a calm environment, and clear integration into existing support plans. Training for professionals, flexible usage options, and short, adaptable sessions were emphasized as essential.

Other studies also confirm that social robots increase motivation for interaction. For example, Van den Berk-Smeekens et al. (2020) and François et al. (2017) developed robot-based interventions based on Pivotal Response Treatment (PRT), which promoted self-initiated interactions. The children responded well to the therapy and showed positive emotional reactions. In the area of perception support, Chen et al. (2020) introduced a robot architecture operating from a first-person perspective-an innovation that can improve social cognition in children with ASD. In the motor domain, the use of the Probo robot showed positive effects on handwriting and social independence, according to Palsbo and Hood-Szivek (2012). No suitable studies were found regarding sensory support.

Robotics can also open new paths in diagnostics. For example, Yun et al. (2017) used a parrot-like robot for screening children aged 2 to 11. The robot served as an interactive interface for early detection of abnormalities.

A systematic literature review in 2023 identified nine suitable studies involving a total of 186 children (up to six years old). The studies included various robots: a) Probo, b) KASPAR, c) NAO, d) RoboParrot, e) Charlie, and f) Cozmo (see figure).

The interventions ranged from storytelling and social games to gesture training. While some studies showed significant improvements, the only randomized controlled trial (RCT) in the sample (Zheng et al., 2020) found that robot-assisted interventions are not equally effective for all children.

The technical capabilities of the robots vary widely: Probo has 20 degrees of freedom in its face, NAO has 25 across the entire body, while Cozmo and Charlie operate with reduced but child-appropriate functionalities. The evaluation of the robots was based on criteria such as mobility, expressiveness, sensory features, and everyday usability. Structured games tailored to the age and individual needs formed the core of most interventions.

Nevertheless, challenges remain: there are often gaps between the expectations of professionals and the actual technical possibilities. Additionally, high costs, insufficient training opportunities, and limited societal acceptance present obstacles. Future developments should focus on cost-effective, low-maintenance robot models, standardized training programs for professionals, and AI-supported personalization.

Overall, current research shows that social robots offer great potential to support children with ASD in their development. It is crucial that their use is individualized, sensitive, and well-considered-always in close collaboration with the people involved.

Sources:

Ghiglino, D., Chevalier, P., Floris, F., Priolo, T., & Wykowska, A. (2021). Follow the white robot: Efficacy of robot-assistive training for children with autism spectrum disorder. Research in Autism Spectrum Disorders, 86, 101822. Source

Gómez-Espinosa, A., Moreno, J. C., & Pérez-de la Cruz, S. (2024). Assisted Robots in Therapies for Children with Autism in Early Childhood. Sensors (Basel, Switzerland), 24(5), 1503. Source

Huijnen, C. A. G. J., Lexis, M. A. S., Jansens, R., & de Witte, L. P. (2017). How to Implement Robots in Interventions for Children with Autism? A Co-creation Study Involving People with Autism, Parents and Professionals. Journal of Autism and Developmental Disorders, 47(10), 3079–3096. Source

Salhi, I., Qbadou, M., Gouraguine, S., Mansouri, K., Lytridis, C., & Kaburlasos, V. (2022). Towards Robot-Assisted Therapy for Children With Autism—The Ontological Knowledge Models and Reinforcement Learning-Based Algorithms. Frontiers in Robotics and AI, 9. Source