Abstract
This paper combines matrix game theory with negotiating theory and uses U-solution to study the framework of the consulting air combat of UAV cluster. The processes to determine the optimal strategy in this paper follow three points: first, the UAV cluster are grouped into fleets; second, the best paring for the joint operations of the fleet member with the enemy fleet members are calculated; thirdly, consultations within the fleet are conducted to discuss the problems of optimal tactic, roles of main/assistance, and situational assessment within the fleet. In order to improve the computing efficiency of the framework, this article explores the use of the NVIDIA graphics processor programmed through MATLAB mixed C++/CUDA toolkit to accelerate the calculations of equations of motion of unmanned aerial vehicles, the prediction of superiority values and U values, computations of consultation, the evaluation of situational assessment and the optimal strategies. The effectiveness evaluation of GPGPU and CPU can be observed by the simulation results. When the number of team air combat is small, the CPU alone has better efficiency; however, when the number of air combat clusters exceeds 6 to 6, the architecture presented in this article can provide higher performance improvements and run faster than optimized CPU-only code.
Highlights
In the 2017 Super Bowl midfielder show, Intel used 300 Intel drones to create a volleyball light show, and issued colorful lights, changed formations, and produced various patterns to employ drones besides aerial shooting or investigation, and ‘performance’
This paper studies cluster UAV cooperative air combat simulations
This paper tests the parallel operation of UAV cluster air combat simulation and tests respectively: parallelizes the CPUs of the equations of motion of the two warring parties, the missiles of the warring parties guide using CPU parallelization, and the CPU decentralization calculations joining CUDA technology
Summary
In the 2017 Super Bowl midfielder show, Intel used 300 Intel drones to create a volleyball light show, and issued colorful lights, changed formations, and produced various patterns to employ drones besides aerial shooting or investigation, and ‘performance’. The expert system [2] is one of the traditional methods of intelligent air combat maneuver decision-making, but it can only be used to solve known problems. The successful application of machine learning technology in many fields has provided new ideas for the study of intelligent air combat decision-making. It is an effective method of applying reinforcement learning (RL) in the field of air combat, because it can interact with the environment through repeated experiments and obtain optimal strategies through iteration [11].
Sign-in/Register to view highlights & summary 
Published Version (
Free)
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call