A hierarchical risk-aware UAV path-planning framework coupling risk-distance weighted A^* and improved dung beetle optimization

Urban unmanned aerial vehicle (UAV) operations face significant risks arising from dense population exposure and vulnerable infrastructure. This paper presents a hierarchical risk-aware path-planning framework, termed RDA^*-IDB, which explicitly couples coarse-resolution risk-aware global guidance with high-resolution local trajectory refinement. First, a voxel-based three-dimensional urban risk map is constructed by integrating fatality risk and property damage risk. At the global planning level, a risk distance weighted A^* planner, denoted as RDA^*, operates on the coarse-resolution risk map to balance flight distance and risk exposure and then generate a sequence of risk-aware global waypoints. Each pair of adjacent global waypoints is subsequently mapped to a corresponding high-resolution local planning domain, within which an improved dung beetle optimizer refines the trajectory under path length, turning angle, and obstacle feasibility considerations. Tent chaotic initialization and Lévy flight perturbation are incorporated to improve population diversity and reduce premature convergence. Simulations are conducted in a 5 km $ \times $ 5 km urban area of Jinan, China, using the flight parameters of a DJI M210 RTK UAV. Under the considered simulation conditions, the proposed framework achieves the lowest unit-path average risk value among the evaluated methods. However, the improvement over A^* is small, and the longer generated path results in a higher accumulated path risk cost than those obtained by A^* and RRT^*. The results therefore reveal a scenario-specific trade-off between average relative risk intensity per unit distance and total path length.