With the advancement of museum digitization, how to enhance user visiting experiences and service efficiency through intelligent means has become a significant research direction. This study focuses on the design of navigation systems in the context of audiovisual integration, aiming to achieve precise navigation and efficient interaction. Personalized recommendations are implemented using collaborative filtering and LSTM models, combined with YOLO and SLAM algorithms for exhibit recognition and indoor positioning. The A* path planning algorithm is utilized to generate optimal visiting routes, while heatmap-based hotspot detection technology is employed to identify peak visiting areas and times. The experimental results show that the F1 score of the recommendation algorithm is 0.67, and the recommendation accuracy rate for some users is more than 0.88. The A* algorithm exhibits good planning efficiency with average running times of 7.5 ms, 21 ms, and 42.5 ms in different scenarios. Heatmap analysis reveals that the high-traffic periods are between 10:00–11:00 and 14:00–15:00, with a maximum dwell time of 450 s for users in Area E. In summary, the study indicates that the audiovisual integrated navigation system effectively enhances visiting efficiency and user experience. However, there is still room for optimization in recommendation performance for users with sparse behavioral data and path planning in complex scenarios. Future work can further refine algorithm models and interaction design to improve the system's response speed and user satisfaction.