Personalized Visiting Route Generation for Virtual Museums Based on Large-Scale Models

Virtual museums, as key platforms for cultural dissemination, struggle with non-personalized visiting routes and limited immersive experiences. Large-scale models, with their natural language processing and image understanding capabilities, offer solutions by enabling intelligent Q&A and image analysis. This paper presents a personalized visiting route generation system for virtual museums based on large-scale models. The system optimizes routes through three main components: (1) sampling point generation and scoring using LVLMs; (2) route generation and smoothing based on user interests; and (3) dynamic route optimization via user feedback and LLMs. It leverages the traveling salesman problem and A* algorithm for efficient route planning, and enhances user interaction through voice input and virtual reality visualization. This paper also designs user experiments to qualitatively and quantitatively demonstrate the effectiveness and convenience of the method. Results show significant improvements in user satisfaction and personalized experience.