In this thesis, we investigate the role of emotions and memory in social robotic companions. In particular, our aim is to study the effect of an emotion and memory model towards sustaining engagement and promoting learning in a long-term interaction. Our Emotion and Memory model was based on how humans create memory under various emotional events/states. The model enabled the robot to create a memory account of user's emotional events during a long-term child-robot interaction. The robot later adapted its behaviour through employing the developed memory in the following interactions with the users. The model also had an autonomous decision-making mechanism based on reinforcement learning to select behaviour according to the user preference measured through user's engagement and learning during the task. The model was implemented on the NAO robot in two different educational setups. Firstly, to promote user's vocabulary learning and secondly, to inform how to calculate area and perimeter of regular and irregular shapes. We also conducted multiple long-term evaluations of our model with children at the primary schools to verify its impact on their social engagement and learning. Our results showed that the behaviour generated based on our model was able to sustain social engagement. Additionally, it also helped children to improve their learning. Overall, the results highlighted the benefits of incorporating memory during child-Robot Interaction for extended periods of time. It promoted personalisation and reflected towards creating a child-robot social relationship in a long-term interaction.
Date of Award | 2018 |
---|
Original language | English |
---|
- human-robot interaction
- robotics
- human factors
- social aspects
An emotion and memory model for social robots : a long-term interaction
Ahmad, M. I. (Author). 2018
Western Sydney University thesis: Doctoral thesis