Why Are You Upset?A Framework for Dynamic Behavior Management in Child-Robot Interactions

Socially assistive robots (SARs) are increasingly present in educational settings, supporting teachers in typical classroom environments by providing individualized attention to students. This paper introduces a novel framework for dynamic behavior management in childrobot interactions (CRI), leveraging Applied Behavior Analysis (ABA) principles to enhance educational robotics. The system features a social robot that (1) learns a model of a child’s goals, beliefs, and intentions, grounded in observations and conversations to clarify the detected behavior, and (2) responds effectively to behaviors by planning appropriate sequences of actions to implement strategies suggested by experts. The proposed architecture integrates real-time behavior monitoring, functional behavior assessment (FBA), and adaptive response planning, enabling socially assistive robots to facilitate individualized learning experiences. By utilizing cloud-based processing and local execution, as well as large language models (LLMs) and Planning Domain Definition Language (PDDL), the robot dynamically adjusts its actions based on the identified purpose of a child’s behavior, providing educational activities, monitoring, and applying behavior management strategies. Results from the experimental evaluations highlight the system’s replanning and cloud response times, along with its overall effectiveness.