Tracing historical paths : intelligent AR guide

Abstract

AI has found its’ place in tourism through many approaches, but has not yet solved the problem of dense amassment from visitors to sites with the most effective advertising. This poor dispersal of visitors is the result of a lack in journey personalisation, where tourists visit the most popular generic locations rather than their tailor made choices. A platform offering tourist assistance is an ideal opportunity for the application of AI, which provides a unique customisable experience without the need for expensive human intervention. By training a search-space that reflects tourist areas with historical sites, a representation of location preferences is interpreted by a recommender and path-finder system. These functions, as well as augmented reality features (AR), adapt this information to assist users based on their explicit preferences, such that it optimises the journey for individuals looking to explore new areas and ensure it is reached through enjoyable means. This thesis introduces this platform including the aforementioned intelligent pathfinding and recommender system with the assistance of AR features. Rather than providing the shortest path, a route is generated such that it optimises the journey to feature the most attractive locations according to user’s preferences. These preferences influence the outcome of the route as well as a recommender system, which provides weighted suggestions with the highest matched options at the top of its list. Precise recommendations have been given by the recommender system with relation to site quality and user preferences. This includes a 99.5% accuracy in its distance measurement of real-to-virtual world scale. AR features are generated to mark destination positions and an AR arrow appears on screen to assist navigation when routing. This however, has been measured to vary between a 15◦ and 165◦ angle away from its real world position. The path-finder system is evaluated through a human expert to assess generated paths in context of directions preferences. While not practical enough to take the place of a human guide, the path finder creates routes which are significant to its search-space to preference values and generates customisable journeys which tourists can enjoy.