Personalizing Mixed Initiative Dance Interactions with a Socially-Aware Robot
Open AccessIn recent years, the field of Human-Robot Interaction (HRI) has seen tremendous advancements in understanding how to design effective interactions between humans and robots. One application of HRI is in the use of Socially Assistive Robots (SARs) to design interactions that aim to help people socially, emotionally, and cognitively through friendly and helpful social behavior. These robots have become increasingly capable of providing support in a number of applications such as educational tutoring, elderly care, and interventions for Autism Spectrum Disorder (ASD). In particular, the application area of ASD interventions offers many opportunities for a positive impact on children’s emotional well-being, social relationships, mental health, and general skill development. This dissertation presents a body of work that seeks to understanding how to effectively use assistive robots to design child-robot interactions that can encourage physical activity and promote social engagement in children with ASD. We conducted several user studies to evaluate some salient aspects of this application, including the design of a social engagement model, the effect of multiple robot roles within a dance interaction, and the impact of personalization of robot role transitioning on physical activity and social engagement. In doing so, we created a child-robot interaction design that implements techniques used in ASD therapy, such as prompting, reinforcement, and time delay and a social engagement model to monitor a child’s behavioral responses. We also created two unique paradigms of robot dance behaviors, an architecture to synthesize robot choreographies, and a music-generation platform that coordinates musical output with full body movements. Lastly, we designed the interaction as a mixed initiative interaction, which is a flexible interaction strategy where humans and computer agents take initiative to contribute to achieving a goal. The mixed initiative design facilitates the allocation of robot roles such that the goal of sustaining child engagement is achieved. The decision-making strategy is implemented as a reinforcement learning problem that uses the real-world interaction experience to allocate robot roles. This led to the development of a Personalized Role Transitioning model that dictates the transitions in robot roles so as to personalize the interaction to the child’s behavioral responses. The work presented here informs the design of a robot-assisted dance-based intervention for children with ASD and contributes to a broader understanding of the role of personalization in child-robot interactions.
- All rights reserved
Notice to Authors
If you are the author of this work and you have any questions about the information on this page, please use the Contact form to get in touch with us.