This study introduces a context‐aware human–robot interaction framework designed for intelligent museum guide robots. The system adopts a three‐layer architecture—perception, understanding, and behavior execution—to facilitate adaptive and meaningful interactions in dynamic museum settings. The perception layer integrates RGB‐D cameras, microphone arrays, and laser rangefinders for comprehensive environmental sensing. An attention‐based model in the understanding layer identifies visitor interest points and infers interaction intent, achieving 87.4% accuracy in interest detection and 83.3% accuracy in overall context recognition. The behavior execution layer implements an “approach–explain–retreat” interaction paradigm responsive to engagement levels and prior interaction history. A field evaluation with 50 participants at the Hubei Provincial Museum demonstrated a 68% reduction in inappropriate interruptions, 42% improvement in information relevance, and 57% decrease in early interaction terminations. Visitor engagement time increased from 3.8 min for first‐time users to 7.3 min for return visitors, and information recall improved from 42% to 73% under the context‐aware model. Overall user satisfaction averaged 4.3/5, with 4.5/5 for educational value. These findings highlight the system's effectiveness in enhancing natural and informative robot–human interaction in cultural spaces, contributing to the development of socially intelligent robotic systems.
Zou et al. (Sun,) studied this question.