Distributed UAV Flocking Control And Simulation Research Based On Behavioral Method

UAV flocking technology has a wide range of application prospects in military and civilian fields,and countries around the world are conducting research in the fields of flocking formation,flocking communication and flocking system sensing,etc.The theoretical research and core technology related to the flocking field has become a hot spot for domestic and international research,so it is important to accelerate the research on flocking-related technology research to achieve a leading position in the future battlefield.At present,some milestones have been achieved in flocking technology,although there is still a long distance from practical applications in distributed algorithm design and flocking formation avoidance.In this study,a distributed flocking formation algorithm is designed from natural phenomena and practical application requirements,aiming to meet the needs of flocking formation and obstacle avoidance in an unknown environment,laying a certain theoretical research foundation for flocking upper-level task implementation and providing relevant technical support.The main research contents of the paper are as follows:(1)Neighborhood model and information flow model are designed with reference to existing models and research on bird flocks in nature,different flocking behaviors are designed based on the neighbor model and information flow model,distributed flocking algorithm and circular formation algorithm are designed based on the basic principles of behavior method,and the two algorithms are implemented in Matlab.The performance of the flocking algorithm and the circular formation algorithm under different scenarios is analyzed,and the influence of the parameters in the algorithm on the overall performance of the flocking is studied to lay the foundation for subsequent algorithm improvement and optimization.The differences between the performance of the algorithms under the custom neighbor model and the traditional neighbor model are also compared,and the superiority of the custom neighbor model is verified by comparing the quality of the flocking formation.(2)The flocking algorithm and circular formation algorithm are studied for the flocking formation problem after obstacle avoidance,and two research methods are innovatively proposed for the optimization of flocking algorithm parameters and algorithm improvement.Based on the GA genetic algorithm,the parameter values in the flocking algorithm are optimized,the fitness function of the algorithm optimization is designed,the optimized values of the four important parameters in the algorithm are obtained,and the changes of the formation before and after the optimization are analyzed to verify the effectiveness of the algorithm optimization.Meanwhile,for the problem of unstable formation after obstacle avoidance,an improved flocking algorithm based on virtual centroid fusion is proposed,the selection of centroids is designed,and the performance of the algorithm on formation quality before and after the improvement is compared and analyzed to verify the necessity and effectiveness of the algorithm improvement.(3)Pre-tracking experiments of a single quadrotor UAV were conducted,and the tracking errors of the single speed and displacement during the actual flight were given to provide the prerequisites for the subsequent simulation experiments.A flocking simulation system based on ROS was built,and the flocking algorithm and circular formation algorithm were simulated and experimented jointly by PX4 self-pilot,GAZEBO simulator and QGC ground station,and the deployment of both algorithms in ROS environment was completed.Through the algorithm formation experiments,we verified the effectiveness of the flocking algorithm and the circular formation algorithm by verifying that the flocking accomplished the overall flight to the target point,the autonomous switching of the algorithm,and the circular formation near the target point under the condition that only the neighbor information is available.