Deep learning enabled localization for UAV autolanding (2018-2021)

Minghui Li1     Tianjiang Hu1,*    

1Machine Intelligence and Collective RObotics (MICRO) Lab, Sun Yat-sen Univeristy, Guangzhou 510275, China

Fig. 1. Schematic diagram of ground stereo system for autonomous landing of the fixed-wing aerial vehicle



Dataset Download

download Air2Land Dataset

Link 1: [Baidu Drive] (Extracting Code: 6tk1)

Link 2: [GitHub]


Detection Method

Fig. 2. Structure of detection model based on BboxLocate-Net and PointRefine-Net

Detection Experimental Results

1) BboxLocate-Net model based detection experiments:

Fig. 3. Test results of different CNN algorithms

2) PointRefine-Net model based detection experiments:


Fig. 4. Comparison of detection results at 6 key points before and after coordinate correction

Fig. 5. Comparison of key point detection results before and after coordinates correction



Localization Experimental Results

1) BboxLocate-Net model based Localization experiments:

Fig. 6. Localization results using different UAV detection algorithms

Fig. 7. Accuracy comparison of spatial positioning results


2) PointRefine-Net model based Localization experiments:


Fig. 8. Comparison of localization results before and after PointRefine-Net

Fig. 9. Comparison of key point detection results before and after PointRefine-Net (PR-N) coordinate correction

Fig. 10. Key point detection results detected by four detection algorithms