Advantages of Robot-Assisted Counseling: Can Caregivers Better Address the Concerns of Children with Autistic Spectrum Disorders Via a Small Humanoid Robot?

Background: Children with ASD may experience difficulties communicating concerns to caregivers (e.g., clinical psychotherapists, special education teachers, etc.) due to deficits in social communication ability. Previous research indicates that children with ASD may be able to communicate effectively with robots; this possibility is partially supported by certain positive response tendencies in non-verbal communication (e.g. Cabibihan et al., 2013). Nonetheless, these tendencies do not confirm whether robots may facilitate therapeutic communication in a counseling context. 

Objectives: Tele-operated robots’ potential to assist caregivers in indirectly counseling children with ASD was examined. We provided caregivers with opportunities to talk with participating children via a small humanoid robot; we subsequently qualitatively and quantitatively analyzed the conversations, which concerned the children’s concerns.

Methods: Three female teenagers (referred to as “A,” “B,” and “C”) with ASD participated. The participants’ caregivers had been unable to address their problems with human relationships. All participants directly and indirectly talked with their caregivers. The conversation times ranged from one to four In the indirect phases, participants talked with a 30-cm-height humanoid robot.. The caregiver could remotely control it to utter the caregiver’s messages using a keyboard interface. We qualitatively and quantitatively examined counseling achievements throughout the series of conversations.

Results: The average total time of conversation was 17.0 and 5.5 minutes in the indirect and direct phases, respectively. The utterance ratio (UR) was calculated as the ratio of normalized time spent on utterances by participants in the indirect phase, compared to in the direct phase. The average UR across all participants was 2.2 (SD = 1.1), indicating that participants uttered more in the indirect phases than in the direct phases. All participants were able to discuss concerns that had not previously been disclosed to their caregivers; however, counseling achievements varied among participants. A and B were able to discuss previously undisclosed worries; specifically, A disclosed the names of people she disliked, and B disclosed circumstances under which she felt uncomfortable interacting with her friends. By contrast, C was able to inhibit excessive aggression toward particular classmates; previously, she had been unable to avoid panicking even when merely encountering these classmates, and no caregiver had been able to adequately manage her panic reaction. C conversed with the caregiver for a particularly long time through the robot regarding these classmates, although she had previously experienced difficultly discussing that topic directly with the caregiver. As she promised to the robot in conversation, she successfully met one of these classmates while keeping calm. Significantly, the caregiver reported that C had been able to keep calm in the presence of these classmates including when she met him two weeks after the experiment.

Conclusions: The caregivers generally succeeded in counseling children with ASD via a small humanoid robot. This result encourages us to develop and introduce a robot-assisted counseling environment into the treatment and education of children with ASD, although the sample size has been limited. Long-term follow-ups and investigation of customization of the technique to reflect individual care needs are important issues for future research.