This methods blog is about my experiences and reflections regarding research I carried out with young children and social robots as part of a four-country (Australia, Belgium, Italy and the UK) study which is investigating young children’s – and their families’ – domestication of an internet-connected robotic toy called Cozmo. The research project has been funded by the Australian Research Council (DP180103922) and is led by Dr Donell Holloway at Edith Cowan University, Perth.
The social robot
Cozmo is what is usually defined as a social robot, that is, a robot designed to interact with humans as a human, thus simulating the affective dimensions of human relationships (Zhao, 2006). More specifically, Cozmo has two humanoid features: eyes and speech. First, it has a small screen provided with eyes, that help it simulate facial expressions. Second, it can speak and sing. It is equipped with three cameras and facial recognition software, but it is not provided with speech recognition software – rather its ability to talk and sing is controlled through the app.
Our ambition within this project was to understand how young children make sense of Cozmo’s embodied agency – the ways it sees, speaks, communicates its emotions, moves in space, learns, etc.; how do they interact with the robot; what play, communicative and learning practices do they engage in; how Cozmo is incorporated within the temporal structures and the digital-material arrangements of domestic life; in a word, the domestication of Cozmo in the home environment.
Theoretically, internet-connected robotic toys are boundary objects: they are simultaneously toys, media and social robots (Mascheroni & Holloway, 2019), and cross the boundaries between digital and non-digital, material and immaterial, online and offline, etc. (Marsh, 2017).
In order to grasp the complex digital-material assemblages that constitute play practices involving children and internet-connected toys, we combined discursive methods, observations, visual methods, and co-construction. We also focussed on ways in which children and parents could meaningfully participate in the research process. More specifically, the first family visit consisted of an interview with parents and a play-tour with the children. At the end of the first visit, children were given a Cozmo and were shown the basic functions by the researchers. In-between visits, parents were asked to film their children’s interactions with Cozmo. After 4-6 weeks, we paid a second visit to the family consisting of; an interview with parents to discuss their impressions and the data (video footage) they had collected; and, a co-construction activity with children – the co-construction of Cozmo’s story with the aid of the Book Creator app.
Reflections and takeaways
Both the play tour and the Book Creator activity were designed to engage children in participatory research through activities that are not only appropriate for their age, but also meaningful and fun. The co-construction of Cozmo’s story through words, videos, images and sounds was particularly helpful, since it helped children engage in reflexivity. In fact, the book creation followed a script that invited children reflect on their interactions with Cozmo: how the robot responded to their words, gestures and commands through the app; how the robot sees and moves in space; how it learned through time, and; what Cozmo’s feelings for them and their feelings for Cozmo were. Therefore, it provided an insight into the meaning-making practices through which children make sense of social robot’s agency and embodied intelligence. For example, it was during the co-creation activity that children explained how they made sense of the times when Cozmo failed to recognise their faces or understand their questions: some wondered whether Cozmo could not recognise them because they had a different hair style or different eyewear, others suggested that Cozmo could not understand them because he doesn’t speak Italian.
The Book Creator activity, though, posed some challenges, as children soon became passionate about the design of the book and paid much attention to choosing the font and the layout. So, instead of the twenty minutes foreseen in the research protocol, the activity lasted 30-40 minutes on average. Despite this unforeseen time issue, the children’s interest in producing a digital book with the researcher about Cozmo (including making time-consuming design decisions) was very apparent and indicative of their role as engaged co-participants in the research process.
The flow of the narration when co-constructing the book was also complicated by Cozmo itself, which, from time to time, demanded to play, either with sounds or through notifications via the app. So, it was not uncommon for children to engage in play sessions during the co-construction activity. Children also positively reacted to the researcher’s invitations to include short videos or pictures of Cozmo in the book. This provided the researchers with further opportunity to observe children’s interactions with Cozmo and collect additional visual material.