Robot-assisted Training for Facilitating Prosocial Behaviors of Children with Autism Spectrum Disorders

本研究的目的是探索个性化机器人辅助训练自闭症谱系障碍儿童的潜力。这项研究探讨了如何使用机器人向自闭症谱系障碍儿童提供反馈。特别是，我们研究了如何在培训期间保持每个孩子的参与。我们探讨了潜在的关键因素，在预测的机器人使用视频观察和生理信号为基础的方法的行为。这项研究发现，自愿的亲社会行为发生在微笑和接近机器人之后。基于这些发现，我们可以通过概率模型检测微笑来预测亲社会行为。此外，我们研究了对机器人的非自愿行为。当孩子转身离开机器人时，父母的口头反馈可以有效地让孩子重新开始互动。这表明，孩子的微笑可能表明与机器人接触的内在动机。相比之下，当孩子没有自我激励与机器人互动时，父母的反馈可能有助于保持参与。

The purpose of this study was to explore the potential of personalized robot-assisted training for children with Autism Spectrum Disorder. This research explored how to provide feedback to individual children with Autism Spectrum Disorders using a robot. In particular, we investigated how to maintain each child's engagement during the training. We explored potential key factors in predicting behaviors toward a robot using video observation and a physiological signal-based method. This research found that voluntary prosocial behaviors occurred after smiling at and being closer to the robot. Based on the findings, we could predict prosocial behavior by detecting smiling with a probabilistic model. Additionally, we investigated involuntary behaviors toward a robot. When a child turned away from the robot, verbal feedback from a parent was effective to make the child reinitiate the interaction. It suggests that the smiling of a child might indicate inner motivation to engage with a robot. In contrast, when a child is not self-motivated to interact with a robot, parental feedback can be helpful to maintain engagement.