UAV-Based Target Search in Maze-Like Terrain

Abstract

Owing to their being less expensive to operate, greater accessibility, and inherent ability to maneuver over tough terrain, unmanned aerial vehicles (UAVs) have become an indispensable tool for rescue missions. However, in the cases where the missions require real-time decision-making without prior maps, existing approaches to UAV path planning still lack efficiency. The problem addressed in the paper is the design and assessment of a hybrid path planning algorithm for UAV-supported search and rescue operations adopted in maze-like, two-and-a-half-dimensional (2.5D) environments. We compared the classical algorithms, such as Random DFS, map-based, A-star, and D, with our suggested hybrid approach, which uses both the Frontier and Greedy algorithms, in Python-based simulations. Through this process, we achieved a total success rate (100%) across all algorithms, but we noticed profound differences in path efficiency, which required a lower time budget and computational resources among them. In particular, the hybrid advice had remarkable benefits over the baseline, such as 22% in Steps-to-Target, 46% in coverage speed, and more than 50% in Total Mission Steps. The only point is that the algorithm preserves its real-time feasibility and that the duration of each step is less than 50 ms. This study concludes that the UAV navigation strategy with hybridized frontier and greedy search is a promising and computationally efficient solution for time-critical SAR missions in maze-like environments, with a strong potential application in time-sensitive search and rescue responses.