Deep Reinforcement-Learning-Based Air-Combat-Maneuver Generation Framework

With the development of unmanned aircraft and artificial intelligence technology, the future of air combat is moving towards unmanned and autonomous direction. In this paper, we introduce a new layered decision framework designed to address the six-degrees-of-freedom (6-DOF) aircraft within-visual-r...

Full description

Saved in:
Bibliographic Details
Main Authors: Junru Mei, Ge Li, Hesong Huang
Format: Article
Language:English
Published: MDPI AG 2024-09-01
Series:Mathematics
Subjects:
Online Access:https://www.mdpi.com/2227-7390/12/19/3020
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850184992615825408
author Junru Mei
Ge Li
Hesong Huang
author_facet Junru Mei
Ge Li
Hesong Huang
author_sort Junru Mei
collection DOAJ
description With the development of unmanned aircraft and artificial intelligence technology, the future of air combat is moving towards unmanned and autonomous direction. In this paper, we introduce a new layered decision framework designed to address the six-degrees-of-freedom (6-DOF) aircraft within-visual-range (WVR) air-combat challenge. The decision-making process is divided into two layers, each of which is addressed separately using reinforcement learning (RL). The upper layer is the combat policy, which determines maneuvering instructions based on the current combat situation (such as altitude, speed, and attitude). The lower layer control policy then uses these commands to calculate the input signals from various parts of the aircraft (aileron, elevator, rudder, and throttle). Among them, the control policy is modeled as a Markov decision framework, and the combat policy is modeled as a partially observable Markov decision framework. We describe the two-layer training method in detail. For the control policy, we designed rewards based on expert knowledge to accurately and stably complete autonomous driving tasks. At the same time, for combat policy, we introduce a self-game-based course learning, allowing the agent to play against historical policies during training to improve performance. The experimental results show that the operational success rate of the proposed method against the game theory baseline reaches 85.7%. Efficiency was also outstanding, with an average 13.6% reduction in training time compared to the RL baseline.
format Article
id doaj-art-6d44eb78b0f549c28bac8194d2fce41b
institution OA Journals
issn 2227-7390
language English
publishDate 2024-09-01
publisher MDPI AG
record_format Article
series Mathematics
spelling doaj-art-6d44eb78b0f549c28bac8194d2fce41b2025-08-20T02:16:54ZengMDPI AGMathematics2227-73902024-09-011219302010.3390/math12193020Deep Reinforcement-Learning-Based Air-Combat-Maneuver Generation FrameworkJunru Mei0Ge Li1Hesong Huang2College of Systems Engineering, National University of Defense Technology, Changsha 410073, ChinaCollege of Systems Engineering, National University of Defense Technology, Changsha 410073, ChinaCollege of Systems Engineering, National University of Defense Technology, Changsha 410073, ChinaWith the development of unmanned aircraft and artificial intelligence technology, the future of air combat is moving towards unmanned and autonomous direction. In this paper, we introduce a new layered decision framework designed to address the six-degrees-of-freedom (6-DOF) aircraft within-visual-range (WVR) air-combat challenge. The decision-making process is divided into two layers, each of which is addressed separately using reinforcement learning (RL). The upper layer is the combat policy, which determines maneuvering instructions based on the current combat situation (such as altitude, speed, and attitude). The lower layer control policy then uses these commands to calculate the input signals from various parts of the aircraft (aileron, elevator, rudder, and throttle). Among them, the control policy is modeled as a Markov decision framework, and the combat policy is modeled as a partially observable Markov decision framework. We describe the two-layer training method in detail. For the control policy, we designed rewards based on expert knowledge to accurately and stably complete autonomous driving tasks. At the same time, for combat policy, we introduce a self-game-based course learning, allowing the agent to play against historical policies during training to improve performance. The experimental results show that the operational success rate of the proposed method against the game theory baseline reaches 85.7%. Efficiency was also outstanding, with an average 13.6% reduction in training time compared to the RL baseline.https://www.mdpi.com/2227-7390/12/19/3020air combatdeep reinforcement learningSACrecurrent neural network
spellingShingle Junru Mei
Ge Li
Hesong Huang
Deep Reinforcement-Learning-Based Air-Combat-Maneuver Generation Framework
Mathematics
air combat
deep reinforcement learning
SAC
recurrent neural network
title Deep Reinforcement-Learning-Based Air-Combat-Maneuver Generation Framework
title_full Deep Reinforcement-Learning-Based Air-Combat-Maneuver Generation Framework
title_fullStr Deep Reinforcement-Learning-Based Air-Combat-Maneuver Generation Framework
title_full_unstemmed Deep Reinforcement-Learning-Based Air-Combat-Maneuver Generation Framework
title_short Deep Reinforcement-Learning-Based Air-Combat-Maneuver Generation Framework
title_sort deep reinforcement learning based air combat maneuver generation framework
topic air combat
deep reinforcement learning
SAC
recurrent neural network
url https://www.mdpi.com/2227-7390/12/19/3020
work_keys_str_mv AT junrumei deepreinforcementlearningbasedaircombatmaneuvergenerationframework
AT geli deepreinforcementlearningbasedaircombatmaneuvergenerationframework
AT hesonghuang deepreinforcementlearningbasedaircombatmaneuvergenerationframework