Jumping AI for Unreal Engine

• Autor(a) principal: Silva, Gabriel Quaresma Moreira da
• Data de Publicação: 2019
• Tipo de documento: Dissertação
• Idioma: eng
• Título da fonte: Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos)
• Texto Completo: http://hdl.handle.net/10400.8/4545
• topic: Pathfinding; Jumping AI; Path planning on grids; Voxel based worlds; Domínio/Área Científica::Engenharia e Tecnologia::Engenharia Eletrotécnica, Eletrónica e Informática

Pathfinding plays a vital role in video games, whether in terms of gameplay mechanics or player immersion. Commonly used methods only allow the simplest types of movements like walking and running. Although seldom, other types of movement, like swimming and flying, are also considered. Even rarer are mechanisms that natively contemplate jumps without the need of extra intervention of game developers. Most games overlook these movements on Non Player Characters, decreasing the realism of the experience. This dissertation discusses the limitations of Navigation Meshes when it comes to take jumps into consideration, while offering solutions to some of its problems. However, found solutions lack in automaticity, requiring high implementation times. In the interest of improving upon this problem, a new solution using grid-based any-angle pathfinding is proposed. In this approach, each cell of this navigation grid constitutes a voxel that delimits a small 3D space and is expressed in a shape of a cube. Voxels discretize the game world and are explored by a search algorithm to achieve pathfinding with jumps. In this context, performance is critical and the paths should be optimal and efficient. Results show that the voxel based solution can be successfully applied in game development and that it has relevant characteristics that could justify choosing this method over the navigation meshes alternatives for jumping.