Path Planning for Fully Autonomous UAVs-A Taxonomic Review and Future Perspectives

Abstract

Autonomous Unmanned Aerial Vehicles (UAVs) rely on advanced path planning to operate independently, especially in unfamiliar settings without human intervention. The process typically involves localization, mapping, optimal path selection, motion planning, and control. Achieving autonomous navigation from one point to another requires balancing various factors, such as energy efficiency, speed, cost, path length, and computation time. In this study, a novel taxonomy that categorizes UAV path planning into four distinct classifications is introduced. A systematic review of the technological advancements in path planning methodologies, tracking the evolution from classical techniques to cutting-edge solutions, with a particular focus on dynamic environments is also represented. This review includes a comparative analysis of various methods based on factors including approach utilized, testing platform, environment type, time efficiency, etc. Various relevant parameters and benchmark datasets that are crucial for UAV path planning are also explored. Despite widespread use, current methodologies still face significant challenges, such as handling unknown threats in dynamic environments, effective obstacle avoidance, limited payload capacity, real-time responsiveness, and energy consumption. These issues limit their overall usefulness. This study aims to highlight these challenges and suggest potential directions for future research, ultimately contributing to the advancement of real-time UAV path planning.

Keywords

• Path planning
• deep learning
• deep reinforcement learning
• heuristic approaches
• machine learning
• unmanned aerial vehicle (UAV)