Unmanned Aerial Vehicles (UAVs) have demonstrated their value in disaster response by providing efficient and rapid data collection capabilities. This study focuses on optimizing UAV task allocation and coordination by leveraging deep learning algorithms and Received Signal Strength Indicator (RSSI) signals. By integrating Euclidean distance, centroid algorithms, and Gated Recurrent Units (GRUs), the research proposes a method that enhances localization performance. The simulation results reveal that the GRU model achieved the lowest Mean Square Error (MSE) of 65E−04. This performance surpasses that of Convolutional Neural Networks (CNNs), Long Short Term Memory networks (LSTMs), Transformer Networks, and Graph Neural Networks (GNNs). These findings underscore the proposed method’s significant potential to enhance UAV operations in dynamic disaster environments. These results demonstrate the considerable promise of the proposed method in improving UAV operations in dynamic disaster scenarios. Future studies will focus on integrating new deep learning architectures with diverse models and datasets to further advance UAV task allocation.