This paper presents an algorithm for multi-UAV path planning in scenarios with heterogeneous Global Navigation Satellite Systems (GNSS) coverage. In these environments, cooperative strategies can be effectively exploited when flying in GNSS-challenging conditions, e.g., natural/urban canyons, while the different UAVs can fly as independent systems in the absence of navigation issues (i.e., open sky conditions). These different flight environments are taken into account at path planning level, obtaining a distributed multi-UAV system that autonomously reconfigures itself based on mission needs. Path planning, formulated as a vehicle routing problem, aims at defining smooth and flyable polynomial trajectories, whose time of flight is estimated to guarantee coexistence of different UAVs at the same challenging area. The algorithm is tested in a simulation environment directly derived from a real-world 3D scenario, for variable number of UAVs and waypoints. Its solution and computational cost are compared with optimal planning methods. Results show that the computational burden is almost unaffected by the number of UAVs, and it is compatible with near real time implementation even for a relatively large number of waypoints. The provided solution takes full advantage from the available flight resources, reducing mission time for a given set of waypoints and for increasing UAV number.
A Deep Learning Trained by Genetic Algorithm to Improve the Efficiency of Path Planning for Data Collection with Multi-UAV
