Research On Multi-Task Planning Method For Post-Disaster Rescue Based On Multi-UAV

With the innovative development of computer technology,as a class of intelligent aircraft with excellent characteristics such as high mobility,functional diversity,and high autonomy,unmanned aerial vehicles(UAV)have been widely used in military and civilian fields.Due to the integration of many different types of sensors,UAV can replace manned aircraft or humans to perform many hidden and dangerous tasks.However,despite the increasing degree of integration of UAV,it is still difficult to deploy a single UAV to perform a large number of tasks in a wide range due to its increasingly complex application environment,increasingly diverse task types and its own limitations such as maximum flight time and maximum load.Although the existing related research has mentioned the endurance of a single UAV,it is rarely considered in the problem model,and the distribution characteristics of task points are hardly considered.In addition,the existing literature focuses more on homogeneous UAV and static task planning,and there are relatively few studies on heterogeneous UAV and dynamic multi-task planning.In order to solve the above problems,this thesis based on multi-UAV,with the background of post-earthquake rescue,research on UAV multi-task planning for both static environment and dynamic environment.The main research contents of this thesis are as follows:(1)Aiming at the multi-UAV multi-task planning problem in a static environment,considering the endurance of a single UAV and the distribution characteristics of multi-task points,with the goal of minimizing the overall task execution time,this thesis proposes a Static Multi-UAV Multi-task Planning Algorithm Based on Pre-partition and Repartition Strategy(SMMP-PR).Pre-partition can be used to narrow the scope of a single UAV to perform tasks,and after pre-partition,the UAV does not need to consider its own endurance time limit,and can achieve the goal of optimizing its task execution time on the premise of successfully completing all tasks.Finally,a grouping strategy is used to evenly arrange task areas for multiple UAVs to optimize the overall timeconsuming.(2)Aiming at the multi-UAV multi-task planning problem in a dynamic environment,according to the dynamic change characteristics of UAVs and task points,with the goal of maximizing the overall task completion rate,this thesis proposes a Dynamic Multi-UAV Multi-task Planning Algorithm Based on Pre-assignment,Re-assignment and Task Sequence Adjustment Strategy(DMMP-PR-TSA).The pre-assignment strategy based on the improved Self-Organizing Maps(SOM)enables the initial tasks to be executed to be assigned to suitable UAVs according to the location,type and required resources,and when the environment or UAV state changes,the re-assignment mechanism is triggered to match new UAV for the task.In addition,for the problem that the follow-up task sequence of UAV is changed due to re-assignment,this thesis uses reinforcement learning(RL)method to dynamically adjust the follow-up task sequence,so as to update the follow-up track of UAV and improve the overall task completion rate.(3)Through experimental design,analysis and comparison,the results show that in homogeneous UAV and static multi-task planning problems,SMMP-PR can greatly shorten the overall task execution time;in heterogeneous UAV and dynamic multi-task planning problem,DMMPPR-TSA can effectively improve the overall task completion rate,which is better than the comparison algorithms.