Accurate and reliable positional estimation remains a major challenge for autonomous UAV navigation, especially in complex operational environments where GPS signals are compromised by intentional jamming, spoofing, or intrinsic measurement errors. Furthermore, inertial navigation systems are prone to cumulative drift, which leads to diverging localization inaccuracies over time. This paper presents a visual-aided method for enhancing UAV localization by employing a sequential image correlation algorithm. The proposed approach estimates the relative motion of the observer by matching consecutive image frames, thereby leveraging a cumulative analysis of translation and rotation parameters. The algorithm has been implemented in MATLAB and validated on both real and synthetic datasets. Synthetic data were generated using the Blender environment, which provided a flexible framework for creating virtual scenes with controllable camera trajectories and sensor parameters. The method is designed to augment conventional GPS/IMU-based positioning systems.
原文链接
分享
引用
