Abstract
This paper presents a distributed multi-UAV planning framework designed for navigating through complex, unknown environments. We introduce a motion primitive-based D* path planning algorithm that incorporates the dynamic considerations of UAVs, facilitating further optimization of the planned paths. Each UAV utilizes this proposed algorithm to discover collision-free, kinodynamically feasible, and minimum-cost guiding paths. Subsequently, trajectory optimization is performed to derive an optimized trajectory. This optimization process leverages the convex hull property of B-splines, the gradient of the Euclidean distance field (EDF), dynamically feasible constraints, and the shared trajectories of other UAVs. Additionally, to address trajectories with high collision risk, we employ an adjustment strategy to allocate the knot span of the non-uniform B-spline trajectory. Simulations and real-world experiments validate the effectiveness of our proposed method.

This publication has 6 references indexed in Scilit: