Balancing and transposition of maps for location-based games

• Autor(a) principal: Silva, Luís Fernando Maia Santos
• Data de Publicação: 2019
• Tipo de documento: Tese
• Idioma: eng
• Título da fonte: Repositório Institucional da Universidade Federal do Ceará (UFC)
• Texto Completo: http://www.repositorio.ufc.br/handle/riufc/47059
• topic: Location-based Games; Procedural Content Generation; Graph Isomorphism

Location-Based Games (LBGs) rely on the player’s location to change its game state, usually as the main trait of playability. Thus, developing worldwide LBGs is a challenging task dueto the need to deploy game instances in multiple locations, while maintaining the same game balancing, features, and even correlations between locations of the game and the real world. Since LBGs rely on players’ location, it is virtually impossible to manually design interactions, challenges, and game scenarios for every place a player is at. Therefore, the same LBG is likely to have distinct instances with varying difficulty levels because of differences in terrain, distance, transport availability, etc. As a result, even established game companies struggle to deploy LBGs around the globe, so the current generation of LBGs is not available in many areas, especially small cities and poor neighborhoods of big cities. Additionally, modern LBGs still present huge balancing differences between regions and avoid exploring the competition between players like other game genres. In this thesis, we propose a method for transposing LBGs maps while focusing on maintaining their game balancing. This approach depends on information about Points-of-Interest (POIs) around the players’ location and estimations about the cost to move between POIs. We introduced two measurements to estimate game balancing in modern LBGs and implemented three different algorithms that aim at transposing LBGs’ maps with minimal variations in game balancing. The first measurement, called Internal Balancing Difference, assesses game balancing internally and the second, called Minimum Balancing Difference, compares game balancing between two instances of a game. The transposition algorithms are based on the Monte Carlo tree search, the Ullmann’s algorithm, and Genetic Algorithms. In this case, we convert LBGs into directed weighted graphs and use one of the algorithms to generate an LBG instance according to the player’s location. To validate the proposed approach, we designed four LBGs with distinct features, gameplay, and mechanics, and conducted an experiment that required samples to compare maps generated by these algorithmsin different locations. Results indicate that games with similar game balancing score higher and that the algorithms differ in performance depending on the number of POIs. Finally, we can conclude that this work contributes to improve the development of LBGs by helping to mitigate the challenge of transposing LBGs while maintaining game balancing.