Robotic - Cognitive Adaptive System for Teaching and Learning (R-CASTLE)

Abstract

One of the biggest current challenges in education is to positively impact the teaching-learning process with technology. Two common reasons are the teachers’ lack of preparation and the students’ attention span. Several Human-Robot Interaction (HRI) studies are approaching these issues. However, very few of them are considering both teachers and students in a one and only application. Thus, the presented thesis had two objectives: to provide a unique and intuitive HRI tool for education and to evaluate its impact on the users. The resulting architecture was a Robotic - Cognitive Adaptive System for Teaching and Learning (R-CASTLE). R-CASTLE aims to provide customized interactions and personalized learning to the students through machine learning for autonomous vision and dialogue interactions. The methods are configured by the teachers in the windows of the system’s graphical interface. Teachers can also have access to the system’s evaluations in chart mode of the students’ collective and individual performances. In end-to-end experiments, teachers and students claimed to experienced a sensitive potential of the system to support them. R-CASTLE was tested with other interactive devices in different applications and the results showed high performances in their activities design optimization. From the best of our knowledge, it is an innovative proposal implemented with collaborative assistance of institutes from Portugal, Italy and Japan. R-CASTLE is currently being adapted to support teachers in their remote classes during COVID-19 pandemic.