Mike Ligthart

Mike Ligthart profile

Title: Shaping the Child-Robot Relationship: Interaction Design Patterns for a Sustainable Interaction

Location: Vrije Universiteit Amsterdam

Date: September 30th, 2022

Abstract: Social robots are an interesting piece of technology, especially for children. Children are immediately drawn to them. The social relationship that starts to form between children and robots has a lot of potential to do good. For example, with the research in this dissertation we aim to contribute to the development of a social robot companion for children with cancer to help them cope with stress. However, once the novelty of the robot wears off after the first few encounters, and children notice it has not really anything to offer, they lose interest. This is one of the major challenges in the human-robot interaction community. We need to equip the robot with the right social abilities to keep children engaged and appropriately foster the child-robot relationship. This dissertation has two parts. In part I we designed and studied robot behaviors that could be used in a distractive intervention. For that purpose the interaction must really be meaningful and social for the children. The meaning is provided through storytelling. We developed six interaction design patterns that enable children to co-decide and co-create parts of the story and coordinate their involvement during the co-creation process. By inviting children to interact with a robot that used these patterns we were able to study the effects on the interaction. The results showed that children paid more attention to a robot that uses the patterns, enjoyed the interaction more, and could recall more about the stories. This tells us that the patterns successfully supported children’s engagement. Equally important, children also feel more agency and competence to co-regulate the interaction when the patterns are used. The patterns contributed to a more social interaction between the child and the robot. In part II we designed and studied robot behaviors that would foster the child-robot relationship. We started with enabling the children to properly get acquainted with the robot. First impressions matter. We learned that how children initiate relationships with people is similar to how children form relationships with robots. Important is that the children and the robot are enabled to share things about themselves to each other. This is called self-disclosure. We developed five interaction design patterns that enable children to comfortably self-disclose to the robot. We invited more children to talk to the robot and get acquainted. Results showed that the patterns provide the children with a clear and consistent way to talk to the robot. We learned that if the robot likes the same things as the child and positively affirms what they shared with it, children feel more similar to the robot and self-disclose more. We furthermore learned that both introverted and extraverted children self-disclose more and more intimately if the robot’s behaviors are a bit toned down. The next step was to successfully enable the robot to have a conversation that spans multiple sessions. That is why we imagined each session as an episode in a serial TV show. The robot and the child are the main actors. The content was co-created with professional writers and it shows. Children were looking forward to talking to the robot each week partly because they wanted to know how the story continued. We developed eight interaction design patterns that store children’s self-disclosure and use it to make subsequent sessions with the robot more personal. We called this the memory-based personalization strategy. In a longitudinal study we found that the memory-based personality strategy successfully communicated to the children that the robot remembers them. As a result they keep feeling close the robot. Furthermore, children remained consistently more willing to continue interacting with a robot using memory-based personalization.

Publications: