CARE: towards customized assistive robot-based education.

This study proposes a novel approach to enhancing the learning experience of elementary school students by integrating Artificial Intelligence (AI) and robotics in education, focusing on personalized and adaptive learning. Unlike existing adaptive and intelligent tutoring systems, which primarily rely on digital platforms, our approach employs a personalized tutor robot to interact with students directly, combining cognitive and emotional assessment to deliver tailored educational experiences. This work extends the current research landscape by integrating real-time facial expression analysis, subjective feedback, and performance metrics to classify students into three categories: Proficient Students (Prof.S), Meeting-Expectations Students (MES), and Developing Students (DVS). These classifications are used to deliver customized learning content, motivational messages, and constructive feedback. The primary research question guiding this study is: Does personalization enhance the effectiveness of a robotic tutor in fostering improved learning outcomes? To address this, the study explores two key aspects: (1) how personalization contributes to a robotic tutor's ability to adapt to individual student needs, thereby enhancing engagement and academic performance, and (2) how the effectiveness of a personalized robotic tutor compares to a human teacher, which serves as a benchmark for evaluating the system's impact. Our study contrasts the personalized robot with a human teacher to highlight the potential of personalization in robotic tutoring within a real-world educational context. While a comparison with a generic, unpersonalized robot could further isolate the impact of personalization, our choice of comparison with a human teacher underscores the broader objective of positioning personalized robotic tutors as viable and impactful educational tools. The robot's AI-powered system, employing the XGBoost algorithm, predicts the student's proficiency level with high accuracy (100%), leveraging factors such as test scores, task completion time, and emotional engagement. Challenges and learning materials are dynamically adjusted to suit each student's needs, with DVS receiving supportive exercises and Prof. S receiving advanced tasks. Our methodology goes beyond existing literature by embedding a fully autonomous robotic system within a classroom setting to assess and enhance learning outcomes. Evaluation through post-diagnostic exams demonstrated that the experimental group of students using the AI-robot system showed a significant improvement rate (approximately 8%) over the control group. These findings highlight the unique contribution of this study to the field of Human-Robot Interaction (HRI) and educational robotics, showcasing how integrating AI and robotics in a real-world learning environment can engage students and improve educational outcomes. By situating our work within the broader context of intelligent tutoring systems and addressing existing gaps, this study provides a unique contribution to the field. It aligns with and builds upon recent advancements, while offering a distinct perspective by incorporating robotics to foster both academic and emotional engagement.