Adaptive multi-granularity graph attention fusion algorithm for autonomous amphibious aerial vehicle route optimization

The navigation of autonomous amphibious aerial vehicles (AAVs) relies on temporal and spatial data, including precise location, distance, and information about obstacles, to ensure effective routing. The high volatility of these data and the fact that they are time-sensitive both contribute to increased navigation difficulty. A Multi-Granular Graph (MGG) Attention-Fusion Algorithm (AFA) has been proposed as a solution to this problem. The purpose of this algorithm is to optimize navigational mode, routing selection, and decision-making techniques by utilizing temporal data. Through the use of node–edge graph learning, the method achieves data fusion by utilizing external parameters to reduce complexity. This eliminates the requirement for substantial local data storage. The suggested method identifies optimal routes with less impediments and reduced errors by recognizing global data through maximal fusion. This allows for the identification of optimal routes. The results of experimental verification, which utilized variations in distance, velocity, and motor speed, demonstrate a 13.26% enhancement in route precision, a 12.93% reduction in complexity, and a 12.77% decrease in route error. Complex interplay between sensor inputs, ambient conditions, and vehicle dynamics make the navigation problem intrinsically nonlinear. The suggested MGG-AFA approach uses data fusion and hierarchical attention-based graph learning to capture these nonlinear interactions.