Multi-UAV Target Localization Based on 3D Object Detection and Visual Fusion

The multi-UAV target localization problem aims to cooperatively estimate the coordinates of targets observed from multiple views or perspectives. To address the visual fusion of multi-UAV target localization, a construction method for target localization models and a fusion strategy for multi-view localization information are proposed. The target localization model is constructed based on the input of 3D object detection results. The multi-view target localization is solved through Kalman filtering, which combines information from multiple views in the current frame and fusion of information from previous and subsequent frames. The effectiveness and robustness of the system are verified on a hardware-in-the-loop simulation platform.