An optimization-based selection approach of landing sites for swarm unmanned aerial vehicles in unknown environments

Abstract

Autonomous selection of landing sites is an important capability of a mission when unmanned aerial vehicles (UAVs) face technical difficulties or adverse weather, especially in unknown environments. In order to build a better decision-making system for swarm UAVs landing, besides terrain safety, the global flight cost among individuals should be also considered during descent. However, the existing methods cannot solve these problems elegantly due to terrain uncertainty and optimization complexity. In this paper, we present an optimization method to tackle this issue based on point cloud integrating the coarse-to-fine manner and swarm intelligence model. Specifically, our method starts with a point cloud preprocessing module for sparse elevation estimation and robust path planning. In the coarse stage, a novel cost map is constructed by extracting only low-level features from the elevation map. Based on the optimum results of the coarse selection, the final landing sites are automatically generated and finely evaluated in terms of multiple 3D terrain factors. Finally, the landing model of swarm-UAVs is treated as an unbalanced assignment problem, which minimizes flight costs by incorporating terrain safety, fuel consumption, and path planning. Experimental evaluations on three different real-world scenario datasets demonstrate the effectiveness of our method in both normal operations and emergency situations.