Enhancing UAV Communication Performance Through GCN-Driven Channel Adaptation

Drone deliveries necessitate advanced technology for real-time obstacle recognition and accurate distance calculation; yet, standard LIDAR sensors are costly and provide major weight-related problems. Drone performance is further affected by the substantial weight increase caused by the high cost of acquiring and integrating LIDAR into drones. For drone applications requiring pre-pixel depth estimate, this project provides an affordable alternative using a GCN Spatio-Temporal Channel Adaptation approach. The efficiency of depth estimate has been confirmed by extensive testing. The disparity map yields the object's distance. This shows how block matching and semi-global block matching algorithms may be used to effectively determine depth information using stereo vision. Our study indicates that incorporating GCN Spatio-Temporal Channel Adaptation systems provides a workable and reasonably priced substitute for LIDAR sensors in drone applications.