The Research Of Automatic Obstacle Avoidance Algorithm On Four-rotor UAV Formation

With the development of unmanned aerial vehicle(UAV)technology and artificial intelligence(AI)technology,four-rotor UAV has widely appeared in people's vision.Because of the low cost and flexible operation,the four-rotor UAV is widely used in all aspects of life,industry and military field.As the tasks gradually become more complex,a single UAV has been unable to meet the needs of diverse tasks,and multi-UAVs formation needs to be studied One of the most important parts of the research is the automatic obstacle avoidance of multi-UAVs formation.In this paper,the multi-four-rotor UAVs formation as the research object,its application scenarios,formation rules and flight avoidance are studiedFirst,the flight principle and dynamic characteristics of the four-rotor UAV are analyzed in this paper.Based on the Newton-Euler equation,the simplified study of the previous study is carried out,and the simplified four-rotor UAV is modeled,on which the appropriate flight controller is selected.Second,this paper studies the application scenarios and development trends of the existing UAV formation in military,scientific research and national economy,and creatively proposes a UAV formation rule which conforms to the application scenario and development trend.There are formation characteristics of formation and formation without formation.A formation control method based on behavior and master-slave control is selected based on the scene and rules.Third,in the behavior control mode based on the formation control method chosen in this paper,the global path planning of the UAV is carried out.In this part,we focus on the improvement of the V-map construction method,and this new method is simulated to verify that the method can obtain a V map path which is safer than the original method under certain time conditions.In the improvement of Dual generation method,we put the center of the circle which tangent to the safety circle of obstacle as the V map path node.Through analysis and simulation of Dijkstra algorithm and A * algorithm,the results show that the A* algorithm is more efficient than Dijkstra algorithm in the path finding.Then,we verified that the A * algorithm can plan a safe and reasonable flight path for multi-UAVs formation flight in the off-line state through simulation,and realize the global obstacle avoidance of multi-UAVs formation.Finally,because the A * algorithm is completely dependent on the environmental information when path planning is carried out,and in general the flight area obstacle is not completely known in advance.So we use the immune network planning algorithm for local obstacle avoidance which makes up for the defects of the A* algorithm.In the Idiotypic Immune Network-Based Planning Algorithm(IINBPA),we defined the antibodies for the UAV's flight attitudes based on the information of the local environment,which is detected by the airborne sensor of four-rotor UAV.And the antigens for the obstacles and targets based on these antibodies are designed.Then,the simulation results of the basic genetic algorithm and IINBPA validated the validity and robustness of IINBPA in the local obstacle avoidance of UAV.