Minimizing Road Curvature Through Parallel Computing in Pathfinding Algorithms
Journal
2025 International Conference Automatics and Informatics (ICAI)
Date Issued
2025-10-09
Author(s)
Bosilkov, Filip
DOI
10.1109/icai67591.2025.11324080
Abstract
This paper proposes a curvature-constrained pathfinding approach that aims to minimize the total sum of the turning angles on road networks. We employ an A* search modified to account for turning-angle costs, rather than just distance. To complement this, we integrate parallelization techniques, by assigning path segments or waypoint pairs to different threads, which accelerates the search, especially in multi-waypoint scenarios. Experimental results show that while segmentation-based parallelization can reduce the total turning cost, overhead can counteract speed gains when only two waypoints are involved, at least on a small map. The findings emphasize the utility of curvature-focused routing and highlight the importance of balancing computational overhead with the practical benefits of parallel computing.