Fast Grid-based Path Finding for Video Games William Lee and Ramon Lawrence University of British Columbia Abstract. Grid-based path finding is required in many video games and virtual worlds to move agents. With both map sizes and the number of agents increasing, it is important to develop path finding algorithms that are efficient in memory and time. In this work, we present an algorithm called DBA* that uses a database of pre-computed paths to reduce the time to solve search problems. When evaluated using benchmark maps from Dragon Age TM , DBA* requires less memory and time for search, and performs less pre-computation than comparable real-time search algorithms. Further, its suboptimality is less than 3%, which is better than the PRA* implementation used in Dragon Age TM . 1 Introduction As games have evolved, their size and complexity has increased. The virtual worlds and maps have become larger as have the number of agents interacting. It is not uncommon for hundreds of agents to be path finding simultaneously, yet the processing time dedi- cated to path finding has not substantially increased. Consequently, game developers are often forced to make compromises on path finding algorithms and spend considerable time tuning and validating algorithm implementations. Variations of A* and PRA* are commonly used in video games [9]. The limitation of these algorithms is that they must plan a complete (but possibly abstract) path be- fore the agent can move. Real-time algorithms such as kNN LRTA* [3] and HCDPS [8] guarantee a constant bound on planning time, but these algorithms often require a considerable amount of pre-computation time and space. In this work, we propose a grid-based path finding algorithm called DBA* that com- bines the real-time constant bound on planning time enabled by using a precomputed database as in HCDPS [8] with abstraction using sectors as used in PRA* [9]. The result is a path finding algorithm that uses less space than previous real-time algorithms while providing better paths than PRA* [9]. DBA* was evaluated on Dragon Age TM maps from [10] with average suboptimality less than 3% and requiring on average less than 200 KB of memory and between 1 and 10 seconds for pre-computation. 2 Background Grid-based path finding is an instance of a heuristic search problem. The algorithms studied in this paper, although potentially adaptable to general heuristic search, are specifically designed and optimized for 2D grid-based path finding. States are vacant