A Multi-Drone Joint Map Building Method Based On Multi-variable Layer Directed Cross Genetic Algorithm

Using multiple collaborative UAVs has more advantages in time-critical search and rescue (SaR) missions because they are able to explore the environment quickly and provide higher redundancy than using a single UAV. The key to multi-UAVs Simultaneous Localization and Mapping (SLAM) is to achieve multi-UAVs map-merging. In this paper, we proposed a multi-UAVs map-merging method based on a multivariate layer directed cross genetic algorithm, and uses this method to realize the real-time map fusion of two UAVs based on 3D LiDAR. This method transforms the grid map mosaic problem into an image matching problem, constructs the corresponding mathematical model, establishes the objective function, and designs an improved genetic algorithm to solve the optimization problem of the objective function. In the experiments, the UAVs maintain a fixed altitude during flight, record multiple segments of LiDAR data, and the experiments use simultaneous playback of the data to simulate the collaborative mapping of dual UAVs. The experimental results show that the method can reliably achieve real-time map stitching with good robustness and fast stitching speed.