Influence map-based pathfinding algorithms in video games

• Autor(a) principal: Adaixo, Michael Carlos Gonçalves
• Data de Publicação: 2014
• Tipo de documento: Dissertação
• Idioma: eng
• Título da fonte: Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos)
• Texto Completo: http://hdl.handle.net/10400.6/5517
• topic: Influence Maps; Inteligencia Artifical; Pathfinding; Spatial Reasoning; Domínio/Área Científica::Engenharia e Tecnologia::Engenharia Eletrotécnica, Eletrónica e Informática

Path search algorithms, i.e., pathfinding algorithms, are used to solve shortest path problems by intelligent agents, ranging from computer games and applications to robotics. Pathfinding is a particular kind of search, in which the objective is to find a path between two nodes. A node is a point in space where an intelligent agent can travel. Moving agents in physical or virtual worlds is a key part of the simulation of intelligent behavior. If a game agent is not able to navigate through its surrounding environment without avoiding obstacles, it does not seem intelligent. Hence the reason why pathfinding is among the core tasks of AI in computer games. Pathfinding algorithms work well with single agents navigating through an environment. In realtime strategy (RTS) games, potential fields (PF) are used for multi-agent navigation in large and dynamic game environments. On the contrary, influence maps are not used in pathfinding. Influence maps are a spatial reasoning technique that helps bots and players to take decisions about the course of the game. Influence map represent game information, e.g., events and faction power distribution, and is ultimately used to provide game agents knowledge to take strategic or tactical decisions. Strategic decisions are based on achieving an overall goal, e.g., capture an enemy location and win the game. Tactical decisions are based on small and precise actions, e.g., where to install a turret, where to hide from the enemy. This dissertation work focuses on a novel path search method, that combines the state-of-theart pathfinding algorithms with influence maps in order to achieve better time performance and less memory space performance as well as more smooth paths in pathfinding.