Research on Distributed Task Allocation and Coordination for Multiple UCAVs Cooperative Mission Control
|School||National University of Defense Science and Technology|
|Course||Control Science and Engineering|
|Keywords||UCAV Cooperative Mission Control Distributed Control Task Allocation Task Coordination Contract Net Generalized Partial Global Planning Plan Merging|
Recently, unmanned combat aerial vehicles (UCAVs) have achieved a prominent development in the world, and have imposed a profound impact on the pattern of modern war. With the increasing complication of the battlefield and the increasing diversity of the military mission, it is almost impossible that a single UCAV solely execute the mission. How to control a UCAVs team as a whole to perform a certain military mission and maximize the total operational utility, has become a crucial problem with significant theoretical value and great practical value.With the development of the autonomous capabilities of UCAVs, distributed control has become a promising research field in multiple UCAVs cooperative control. Aiming to overcome the limitations of centralized control methods, the distributed control architecture and associated methods are designed to solve task allocation and task coordination for multiple UCAVs cooperative mission control in this thesis. In this architecture, individual UCAV is regarded as an autonomous agent under limited centralized control, and multiple UCAVs independently plan and mutually negotiate to realize task allocation and task coordination. The main work and the creative contribution of this thesis are as follows:(1) A multi-UCAV cooperative mission control model is presented. The concepts relevant to cooperative mission control are defined, and two typical constraints among tasks are described, i.e. sequence constraints and facilitate constraints. The elements that influence the cooperative operational utility of multi-UCAV are analyzed, and the mathematical model of multi-UCAV cooperative mission control is established. In order to decrease the complexity of the problem, hierarchical and iterative strategy is applied. Thus, the layered cooperative mission control model in the distributed architecture is proposed, and the mission control problem is decoupled and decomposed into two coherent sub-problems, i.e. task allocation and task coordination, which respectively support the cooperation of multi-UCAV at the task level and the coordination of multi-UCAV at the plan level. According to the mutual influences among UCAVs during the task execution process, task coordination is divided into two layers, i.e. plan coordination level to enhance the advantageous influence, and conflict resolution level to avoid the adverse influence. A hierarchical negotiation model is presented, which can realize distributed task allocation, plan coordination, and conflict resolution in different levels.(2) Distributed multi-UCAV task allocation approach based on the extended contract net is presented. The contract types of contract net are expanded to jump from the local optimum that often appears in the original contract net. Buy-sell contract, swap contract, displacement contract, cluster contract are integrated to work together in the extended contract net. The exchange mechanism of contract net is expanded to increase the efficiency of the original contract net. In the extended contract net, multiple exchanges can take place simultaneously in one auction. The simulation results show that the extended contract net can improve the task allocation effectiveness, and decrease the communication magnitude during the negotiation process. The extended contract net that supports the allocation of the tasks with sequence constraints is presented. Thus, UCAVs can deal with the sequence constraints among the tasks when participating in auction and bid to optimize the allocation result. The extended contract net is modeled with Petri nets, thereafter the correctness of the negotiation process and the feasibility of the negotiation result are analyzed.(3) Distributed multi-UCAV plan coordination approache based on conditional contract mechanism is presented. With a view to the constrains among tasks, the optimal team performance depends on not only the result of task allocation, but also the synchronizing activities or the sequencing activities of the UCAVs whose tasks are correlated. A novel negotiation mechanism, conditional contract mechanism based on utility compensation (CCMUC) is presented, which integrates the negotiation framework of generalized partial global planning (GPGP) and the auction-bidding methodology of contract net. Using CCMUC, a UCAV can coordinate its mission plan and tactical behavior with the other UCAVs which perform correlative tasks. Compared with the passive coordination mode of GPGP, CCMUC uses the utility change as the negotiation intermediary. When a UCAV want to get assistances from other UCAVs to improve the total utility, it can actively request them to join the coordination, and compensate them for the disadvantageous agreements which decrease their own benefit. By means of transferring the compensatory utility, CCMUC can support complex multi-link negotiations among multiple UCAVs. The simulation results show that the plan coordination based on CCMUC can further improve the total operational utility after task allocation, and the performance is better than GPGP.(4) Distributed multi-UCAV conflict resolution approach based on plan merging (PM) is researched. When several UCAVs perform tasks in the same spatial region, the crash problems may appear for the temporal confliction and the spatial confliction among UCAVs at the same time. Hence, PM is applied to solve conflict resolution. In order to overcome the shortage of the original PM that can only finish conflict resolution at the qualitative level, the mission plan of UCAV is quantificationally modeled by temporal constrain networks, so PM is generalize at the quantificational level. After exchanging the quantificational mission plan, UCAVs can quickly detect and resolve the temporal-spatial confliction, and incorporate their individual plans into a coordinated joint plan.(5) The route predicting problem for dynamic task allocation and the route planning problem for task coordination are researched. By improving the path planning method based on probabilistic road maps (PRM), changeable coefficient based PRM (CCPRM) method is presented. The simulation results show that when the battlefield situation has changed, without reconstructing roadmaps, CCPRM method can rapidly plan a serial of new predictive routes for task allocation by updating the risk cost of corresponding route segments. By expanding the path planning method based on distance transform, cost-based distance transform (CDT) method is presented. The simulation results show that, by coordinating the length cost and the risk cost of route according to the plan coordination result, CDT method can generate the optimal route that satisfy temporal coordination request.