Taking Learning Out of Real-Time Heuristic Search for Video-Game Pathfinding

Abstract

Real-time heuristic search algorithms are useful when the amount of time or memory resources are limited or a rapid response time is required. An example of such a problem is pathfinding in video games where numerous units may be simultaneously required to react promptly to player’s commands. Classic real-time heuristic search algorithms cannot be deployed due to their obvious state-revisitation (“scrubbing”). Recent algorithms have improved performance by using a database of pre-computed subgoals. However, a common issue is that the pre-computation time can be large, and there is no guarantee that the pre-computed data adequately covers the search space. In this work, we present a new approach that guarantees coverage by abstracting the search space using the same algorithm that performs the real-time search. It reduces the pre-computation time via the use of dynamic programming. The new approach has a fast move time and eliminates learning and “scrubbing”. Experimental results on maps of millions of cells show significantly faster execution times compared to previous algorithms.