To address the long formation time and total travel distance of large-scale UAV collaborative formations, as well as formation control errors caused by delays and interference, we first used the Kuhn-Munkres algorithm combined with the distributed Starfish optimization algorithm to find the optimal position allocation solution. We then used a second-order consistency method that incorporated prediction delay compensation and a dynamically weighted artificial potential field to rapidly form the target formation. Finally, we deployed this method in a unit box integrating an onboard computer and Pixhawk, and used Gazebo, ROS, and PX4 to build a system for multi-UAV formations to conduct three-dimensional, full-coverage detection of a selected area. Taking eight drones as an example, the total formation time was shortened by 18.71%, and the total distance was reduced by 15.73%. The maximum improvement in speed consistency error reached 93.05%, enhancing the efficiency and stability of formation generation. This approach is suitable for situations where drones need to quickly form a specific formation to cover a large area.
Song et al. (Tue,) studied this question.