Child-Robot Behavioral Alignment and Creativity Performance
Tid: On 2022-10-05 kl 13.00
Ämnesområde: Informations- och kommunikationsteknik
Respondent: Maha Elgarf , Beräkningsvetenskap och beräkningsteknik (CST)
Opponent: Professor Tom Ziemke, Linköping University
Handledare: Christopher Peters, Beräkningsvetenskap och beräkningsteknik (CST)
In recent years, robots have been prevalent in almost all domains. One of the most common applications of social robotics is for education with children. This dissertation addresses the integration of creativity-related education in child-robot interactions. Creativity is a required skill in the 21st century. It is regarded by many researchers as an essential survival skill. It has been established that current educational methods limit children's freedom of expression and therefore, negatively impact their creative abilities. To date, a few research attempts have focused on developing social child-robot interactions to foster children's creativity.
In this work, methods were investigated to boost children's creativity skills through social interactions with a robot in a storytelling context. To define and evaluate creativity, standard four creativity measures were used throughout the thesis: fluency, flexibility, elaboration and originality.
First, a social activity was developed to be performed between a social robot and a child. The activity comprises of two games: an interactive priming game and a storytelling game. The activity has been used throughout the thesis to evaluate implemented algorithms and methods. Second, 3 field studies were conducted with 210 school-aged children (5-10 years old). In these studies, the developed activity was used and notions of emotional alignment and creativity alignment between a child and a social robot were examined. In the context of this work, the concept of behavioral alignment refers to the synchronisation between the robot and the child that results in the child mirroring the robot. Emotional alignment occurs when a child mirrors the robot's emotions. Whereas, creativity alignment results in the child behaving creatively as an effect of interacting with a creative robot. Through the conducted studies, the effects of the various types of child-robot behavioral alignment on children's emotional states, engagement with the robot and children's creativity skills were investigated. Third, a computational model that enables a conversational agent to collaboratively interact with a child in a storytelling activity in a creative manner was produced. The computational model was implemented to be used in an integrated manner with the software interface of the storytelling game. The data collected in the first two studies was used to train the computational model that was assessed through the third and last study.
The findings highlight the effectiveness of social robots in promoting children's creativity skills. They emphasize the potential of the developed educational application (storytelling game interface + computational model) in improving children's creative abilities. This work enriches the literature with new insights on developing robot's behaviors that benefit children's creative processes and therefore, is significant to the child-robot interaction (cHRI) community.