Multi-UAV Coverage Path Planning for Gas Distribution Map Applications

Abstract

Aerial coverage for Gas Distribution Map (GDM) using Unmanned Aerial Vehicles (UAVs) with olfaction capabilities has become a key for effectiveness and cost improvements. In this work, the main contribution is the application of Coverage Path Planning (CPP) approaches adapted to survey gas plumes using an autonomous fleet of UAVs. Indeed, the mission is carried out in four steps. First, a gas plume model was integrated to simulate the environment. Then, based on the sensor’s footprint model and the Voronoï diagram, an exact cellular decomposition method is proposed to generate and distribute measurement points or Points Of Interests (POIs) effectively. To take measurements from all these POIs, two planning approaches were proposed in the third step. One is based on the separation of missions between UAVs using the K-means algorithm principle and modeling each path as a Traveling Salesman Problem (TSP). While the second approach models the global mission for all UAVs as a Vehicle Routing Problem (VRP). The resolution of the two mathematical models is achieved using Genetic Algorithms (GAs) and Simulated Annealing (SA) algorithms, respectively. Eventually, the GDM is reconstituted based on triangulation-based cubic interpolation. The evaluation of the proposed simulation techniques based on comparative analyses shows their efficiency, speed and flexibility. The proposed methods take into account the UAV’s flight autonomy and the desired coverage quality, and they provide solutions for a variety of coverage applications, particularly for GDM.