Abstract:

To satisfy the requirements of instantaneity and accuracy of unmanned aerial vehicle(UAV) dynamic path planning for low-altitude penetration under complex terrain conditions, a generalized opposition-based learning adaptive constrained differential evolution (GOBL-ACDE) algorithm is proposed, which is combined with the non-uniform rational B-spline (NURBS) smoothing strategy to improve the accuracy, efficiency, feasibility and airworthiness of UAV dynamic path planning. Firstly, the model of the planning task is constructed as well as the objective cost and constraint function, and we propose a height conversion method to effectively improve the low altitude penetration ability. Then, the performance of the NURBS smoothing strategy is compared with B-spline interpolation and the Bezier curve. In addition, the diversity, convergence and optimization performance of differential evolution are improved through introducing generalized opposition-based learning, adaptive ranking mutation operators and adaptive trade-off model into the algorithm. Finally, through the comparative simulation in static and dynamic environments, it is shown that the proposed method has high accuracy, strong robustness, terrific dynamic performance and excellent reliability in the multi-threat complex terrain, and is able to plan an accurate, efficient and feasible low-altitude penetration path for UAV.

Key words: dynamic planning, differential evolution, non-uniform rational B-spline (NURBS), threat avoidance, low-altitude penetration