Grid Graph Reduction for Efficient Shortest Pathfinding

Single-pair shortest pathfinding (SP) algorithms are used to identify the path with minimum cost between two vertices in a given graph. However, their time complexity can rapidly increase as graph size grows. In this paper, we propose pattern-based blocking algorithm grid (PBGG) that iteratively blocks or reduces free space do not require exploration. The process is based on neighbors of each vertex and utilizes 3x3 binary pattern matching. 0 ( xmlns:xlink="http://www.w3.org/1999/xlink">I · ⌈| xmlns:xlink="http://www.w3.org/1999/xlink">V |/ xmlns:xlink="http://www.w3.org/1999/xlink">C ⌉), where | number vertices, maximum iterations, parallelized cores. PBGG significantly total computation when utilized preprocess an input before applying existing SP algorithms. It also guarantees if minimum-cost exists original graph, then find at least one same reduced proposed method formulated by convolutions be easily implemented using machine learning platforms, such PyTorch. Experimental results show reduce employed conjunction A* Jump Point Search. On average, times 71% for A 41% Search, compared taken alone.