Research On Distributed Mission Planning For Multiple Unmanned Aerial Vehicles

As the increased use of unmanned aeria l vehic les(UAVs)for military missions,the development of a utonomous decision-making has been motivated to ada pt to complex missions with dynamic and uncerta in.Autonomous mission planning is a hot issue in the fie ld of artific ia l inte lligence,autonomous control and operationa l research,and is also a key factor that determines the ability of a utonomous dec ision-making for UAS.Complex military missions which inc lude UAVs,targets/tasks,and the environme nt are common w ith dynamic and uncerta in.In order to meet the new challenges and new requirements for typical military missions on autonomous dec ision-ma king,task allocation,path pla nning and path follow ing,some key technologies for distributed mission pla nning for multi-UAVs,such as task a llocation,path pla nning,path follow ing and integrated mission planning,are studied in this paper.In order to study the mission pla nning of multi-UAVs by means of mathe matical or numerica l simulation methods,a genera l mode l of mission pla nning is first established based on MILP,considering that the mission space containing multiple UAVs is generally dynamic and w ith uncertainty,the complexity,characteristics and corresponding constraints of the mission planning proble m have been ana lyzed,in the condition where the complexity of the task is quite high(e.g.the sum of entities in the task is quite a lot),the hierarchical decoupled simplification method which decomposes the mission pla nning proble m into two independent sub-problems: task allocation and path planning is proposed.At the same time,the advantages and disadvantages of distributed planning architecture and centralized planning architecture is given,so the distributed one is adopted to complete the assignment of tasks and the generation of trajectory,the n a distribute d task planning mode l is established.In addition,in order to verify the feasibility and superiority of the planning results,the optima lity of planning results and the rapidity of the pla nning process are used as indicators of the given evaluation criteria.Based on the proposed distributed mission pla nning mode l and the hierarchical decouple d simplification method,a distributed task a llocation method based on consensus algorithm is designed in the task allocation leve l,and the coordination and conflict-free between UAVs is completed at the task leve l,which can also effective ly decrease the consumption of communication links in the distribute d planning process.For the aspects of the dynamic,communication link limitations,uncerta inty in the military mission,three improvements of the proposed method are given,first of all,the time variable is added to the optimization of the objective function,in this way the algorithm adapts to the dyna mic characteristic of the mission space;secondly,the local communication link c hain is proposed to guarantee the communication connectivity among the UAVs under the condition that the communication link has limitations.In addition,the Bayesian theore m is adopted to transform the continuous/discrete uncertainty of the mission space into the return va lue of the task in the mission,which makes the algorithm adapt to the uncertainty of the mission space;In the SEAD mission,it is required to a llocate the UAV with reconna issance ability and the UAV with attack ability at the same time,w hich can ensure both the return va lue of the task a nd the premise of minimizing the uncertainty of the mission space,in this way the optima lity of planning results is improved.At the path planning le vel,in order to obta in smooth paths that satisfies the constraints of the UAV's dynamics and the environment,at the same time,to e liminate the "deadlock" w hich is derived from the inherent deviation between the smooth paths of the task allocation le vel w ith the path planning leve l,moreover,to improve the solve efficie ncy of path planning proble m,the path planning method based on the para llel A-Star algorithm a nd D ubins path is proposed in this cha pter,on one hand,the problem of "deadloc k" is eliminated by increasing the waiting circ le of the UAV near the target position.On the other hand,a method considering obstac le avoidance for UAV based on the Dubins path is designed.The proposed path pla nning method can quickly generate smooth paths.In addition,a complete six-degree-of-freedom mode l of fix-w inged UAV is established,and the path follow ing method is implemente d based on the vector fie ld guidance algorithm to verify the feasibility of the planning paths.Fina lly,considering the coupling effect of task allocation and path planning,the feasibility of the integrated mission planning proble m is explored.Based on the given SEAD mission,an integrated mission planning mode l inc luding task a llocation information and path information is established based on graph theory and Dubins Car model.The distributed ge netic algorithm is proposed for the given problem.Genetic operators such as multi-type gene coding methods,initia lization,crossover,and mutation are designed,respective ly.The proposed a lgorithm is verified to adapt to t he relative ly simple mission scenarios(with fewer task entities)through simulation analysis and comparison tests.The feasibility and rapidity of the a lgorithm for solving the integrated mission planning proble m and the advantage of the proposed coupled method compared with the hierarchica l decoupled simplification method are discussed,respectively.This paper mainly explores and improves the relevant technologies of multi-UAVs mission pla nning.The re levant research results can provide effective reference for the mission decision-making,task a llocation,and path planning of multi-UAVs,and they can be used to promote the use of unmanned aeria l vehic les in complex military missions,w hich has importa nt the oretical and research significance,together w ith engineering application value.