Swarm Navigation Based on Smoothed Particle Hydrodynamics in Complex Obstacle Environments

In this letter, we propose a method for the navigation of swarm unmanned aerial vehicles (UAVs) in complex environments with obstacles. We propose an algorithmic framework based on Smoothed Particle Hydrodynamics (SPH). In this framework, each UAV is considered a particle, computing its motion information through local interactions with surrounding particles. Based on SPH, the UAV swarm can interactively adjust itself, allowing the entire cluster to advance in the flow pattern of an incompressible fluid. We introduce the Euclidean Signed Distance Field (ESDF) as a representation of the environment. The ESDF is constructed based on the obstacle information in the environment, enabling the swarm to deform and avoid obstacles within the environment. Simultaneously, we propose a swarm navigation function based on B-splines, rapidly obtaining executable trajectories by solving an unconstrained gradient optimization problem. Compared with existing methods, our algorithm exhibits significant improvements in success rate, stability, and scalability. Extensive simulations and physical experiments in both 2D and 3D environments have demonstrated the effectiveness of the proposed method.