Providing the context for project work, coordinator Dr Costas Tzafestas notes: “A prerequisite to achieve successful communication is being able to decode the cognitive state of people around us (intention-reading) and to build trust.” BabyRobot therefore followed a human communication paradigm to model human-robot interaction as a three-step process: sharing attention, establishing common ground, and forming shared goals. Specifically, the research team developed new software modules able to process and analyse multimodal signals from a variety of audiovisual sensors. These sensors capture actions and events in a human-robot interaction environment. They have evidenced superior performance in, among other areas, modelling and recognition of human social, cognitive and affective states, particularly in children interacting with robots.
Engaging with children
Robot learning mechanisms and algorithms developed by one BabyRobot research team enable a robot to adapt its actions and reactions to maximise child engagement within an interactive scenario. “This enhances the potential effect that such a setting may have in ameliorating specific (targeted) cognitive and social (communicative and collaborative) skills in the child,” Dr Tzafestas explains. This development is particularly significant considering BabyRobot’s main target population: children with autism spectrum conditions (ASCs). Project research also included typically developing (TD) children aged 6-10 years. All children interacted in their native language, specifically English, Danish, Greek or Swedish.
Child-robot interaction (CRI) studies
BabyRobot work centred on three CRI use cases. The first covered natural CRI scenarios and the second investigated communication skills development and learning through games. The third use case focused on collaboration skills development and learning via dyadic and triadic interaction, using the robot as a mediator. “These studies have been conducted in parallel by several BabyRobot partners and involved in total over 40 children with ASC symptoms as well as over 150 TD children”, Dr Tzafestas reports. A common goal of these studies was to assess a variety of cognitive and socio-affective skills especially in children with ASCs. They also aimed to evaluate the potential impact that child-robot interactive game scenarios (such as those designed and deployed in BabyRobot) may have in ameliorating such skills in this target population. Project results are very promising and show the great potential of these novel technologies. “The potential societal impact of developing such technologies in the general field of social robotics and further exploiting them in deploying effective child-robot interactive and collaborative game scenarios is presumably immense,” the coordinator states.
Results inspire the way forward
Beyond the science, Dr Tzafestas speaks to the motivation underlying the essence of this research. “The way the behaviour of these children evolved over time, demonstrating for instance signs of empathy towards the robot … or gradually applying collaboration skills, is our research inspiration and drives our way forward,” he asserts.
What are your next steps? How will you move forward?
Future work is aimed at further developing and evaluating BabyRobot technologies and modules. At the same time, several research questions remain open, in particular those related to better understanding the mechanisms of attention and learning in multi-party interaction. “Establishing a CRI framework capable of efficiently manipulating such mechanisms may enable us to effectively address different aspects of learning deficits in children,” Dr Tzafestas concludes.
BabyRobot, robot, children, skills development, communication, collaboration, robot learning, human-robot interaction, child-robot interaction, autism spectrum