TY - JOUR AU - Hu, Yue AU - Harabor, Daniel AU - Qin, Long AU - Yin, Quanjun AU - Hu, Cong PY - 2021/05/25 Y2 - 2024/03/28 TI - Improving the Combination of JPS and Geometric Containers JF - Proceedings of the International Conference on Automated Planning and Scheduling JA - ICAPS VL - 29 IS - 1 SE - Main Track DO - 10.1609/icaps.v29i1.3478 UR - https://ojs.aaai.org/index.php/ICAPS/article/view/3478 SP - 209-213 AB - <p>The JPS family of grid-based pathfinding algorithms can be improved with preprocessing methods such as Geometric Containers. However, such enhancements require a Dijkstra search for every node in the grid and the space and time costs of all this additional computation can be prohibitive. In this work we consider an alternative approach where we run Dijkstra only from every node where a jump point is located. We also compute and store geometric containers only for those outgoing edges which are consistent with the diagonal-first ordering in JPS. Since the number of jump points on a grid is usually much smaller than the total number of grid cells, we can save up to orders of magnitude of time and space. In addition to improving preprocessing overheads, we also present a partial expansion strategy which can improve the performance of online search by reducing the total number of operations on the open list.</p> ER -