Viral Infection Genetic Algorithm with Dynamic Infectability for Pathfinding in a Tower Defense Game
Gabriel Teixeira Galam, Tiago Pereira Remedio, Maurício A. Dias · 2019
Hardware designed for gaming computers and platforms is achieving impressive results on different benchmarks suggesting that the response time of the algorithms have to be improved in order to achieve current real-time requirements. Also important is to consider to maintain the high standard of algorithms results. Pathfinding is an important and computationally complex feature of Non-Player Characters in several video game genres and is still being solved with wellknown A* and Dijkstral algorithms or variations. The problem is that both techniques depend on specific design choices as using a graph as a map for Dijkstra or static environments for the classic A*. Due to good results achieved using theses classical solutions researchers are inclined to include these algorithms in their proposed variations even when a dynamic environment is considered. As a solution this work presents a viral infection genetic algorithm for pathfinding that is able to provide new paths for dynamic changes in the environment and also considering the final amount of the NPC life. Results showed that the proposed algorithm is able to work in real-time obtaining acceptable paths as results.