Pathfinding and Abstraction with Dynamic Terrain Costs


  • Nathan R. Sturtevant University of Alberta
  • Devon Sigurdson Improbable Canada Inc.
  • Bjorn Taylor Improbable Canada Inc.
  • Tim Gibson Improbable Canada Inc.


Abstraction and refinement is a common approach used in games to improve the speed of pathfinding by planning in an abstract space and then refining abstract paths to traversable paths. While there are many variants of this approach that have been developed and studied, research on this problem has largely ignored the problem of pathfinding with terrain types, terrain costs, and dynamic terrain. This paper studies the problem of pathfinding in domains with terrain costs and proposes an abstraction approach that is built around handling terrain costs and dynamic terrain. The resulting approach is able to handle costs in a way that existing approaches do not, and provides a good balance between memory usage, path quality, and pathfinding speed.




How to Cite

Sturtevant, N. R., Sigurdson, D., Taylor, B., & Gibson, T. (2019). Pathfinding and Abstraction with Dynamic Terrain Costs. Proceedings of the AAAI Conference on Artificial Intelligence and Interactive Digital Entertainment, 15(1), 80-86. Retrieved from