Multimodal Interactive Experience Optimization of Museum AR Navigation System: Via 3D Affective Space Model–Integrated Improved A-Star Algorithm

Faced with the increasing demand for museum visitors’ emotional experience, this study constructed a three-dimensional affective space model. Real-time emotional state recognition is achieved by integrating multimodal data sources, including facial expression analysis, speech emotion recognition, and physiological signal monitoring. An improved A-Star (A*) algorithm is proposed by incorporating dynamic weight coefficients into the evaluation function and optimizing the sub-node selection strategy which is applied to interactive path planning in museum AR navigation systems. Experimentally evaluated on a test platform deployed in the Shaanxi Archaeology Museum. The results show that the proposed method reduces average response time by 18.7% (from 3.2 s to 2.6 s), increases emotional recognition accuracy to 89.2%, and improves the path optimization rate by 15.3%. The integration of the affective space model and improved A* algorithm greatly enhances the multimodal interaction quality of the museum virtual navigation interface and strengthens user immersion.