Making social robots adaptable and to some extent educable by a marketplace for the selection and adjustment of different interaction characters living inside a single robot

Abstract

The increasing integration of autonomous robotic systems across various industries necessitates adaptable social interaction capabilities. This paper presents a novel software architecture for socially adaptable robots, emphasizing simplicity, domain independence, and user influence on robotic behaviour. The architecture leverages a marketplace-based agent selection system to dynamically adapt social interaction patterns to diverse users and scenarios. Implemented using ROS2, the framework comprises four core components: scene analysis, a bidding platform, social agents, and a feedback service. A Validation through simulated experiments shows the architecture's feasibility and adaptability, with respect to varying feedback conditions and learning rates. This work lays the foundation for scalable, adaptable, and user-friendly robotic systems, addressing key challenges in industrial and social robotics. Future improvements include enhanced scene analysis, integration of machine learning techniques, and support for more complex behavioural scripts.