Unpacking ORB-SLAM for UAV Navigation and Evaluating Its Efficiency Across Diverse Environments: A Systematic Review

Abstract

Vision-based Simultaneous Localization and Mapping (VSLAM) has become increasingly important for autonomous navigation of Unmanned Aerial Vehicles (UAVs), as it enables simultaneous localization and mapping of unknown environments using visual sensing alone. Among existing VSLAM approaches, ORB-SLAM is distinguished by its robustness under diverse operating conditions, real-time capability, and computational efficiency. This study provides a systematic and comprehensive review of the ORB-SLAM framework, with emphasis on its application in UAV navigation and mapping. We present a detailed analysis of the algorithm’s core components, namely tracking, local mapping, and loop closure. In addition, the challenges associated with deploying ORB-SLAM on aerial platforms are examined, including high dynamic motion, environmental variability, and limited onboard computational resources. This work further offers a comparative evaluation of ORB-SLAM with other SLAM methodologies, highlighting its advantages and limitations across its successive variants. Finally, potential future research directions are discussed to address existing challenges, incorporate deep learning techniques, and evaluate real-world deployment feasibility. This review aims to clarify the role of ORB-SLAM in UAV applications and outline prospective development pathways.