Adaptive Missile Avoidance Algorithm for UAV Based on Multi-Head Attention Mechanism and Dual Population Confrontation Game

In recent years, UAVs have faced increasingly severe and diversified missile threats. To address the challenge that reinforcement learning-based missile evasion algorithms struggle to adapt to various unknown missile types, we introduce a risk-sensitive PPO algorithm and propose a training framework...

Full description

Saved in:
Bibliographic Details
Main Authors: Cheng Zhang, Junhao Song, Chengyang Tao, Zitao Su, Zhiqiang Xu, Weijia Feng, Zhaoxiang Zhang, Yuelei Xu
Format: Article
Language:English
Published: MDPI AG 2025-05-01
Series:Drones
Subjects:
Online Access:https://www.mdpi.com/2504-446X/9/5/382
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849711282411798528
author Cheng Zhang
Junhao Song
Chengyang Tao
Zitao Su
Zhiqiang Xu
Weijia Feng
Zhaoxiang Zhang
Yuelei Xu
author_facet Cheng Zhang
Junhao Song
Chengyang Tao
Zitao Su
Zhiqiang Xu
Weijia Feng
Zhaoxiang Zhang
Yuelei Xu
author_sort Cheng Zhang
collection DOAJ
description In recent years, UAVs have faced increasingly severe and diversified missile threats. To address the challenge that reinforcement learning-based missile evasion algorithms struggle to adapt to various unknown missile types, we introduce a risk-sensitive PPO algorithm and propose a training framework incorporating multi-head attention mechanisms and dual-population adversarial training. The multi-head attention mechanism enables the policy network to extract latent features such as missile guidance laws from state sequences, while the dual-population adversarial approach ensures policy diversity and robustness. Compared to conventional self-play methods and GRU-based evasion strategies, our method demonstrates superior training efficiency and generates evasion policies with better adaptability to different missile types.
format Article
id doaj-art-8fdd7f00794046809b64d5eae7865b29
institution DOAJ
issn 2504-446X
language English
publishDate 2025-05-01
publisher MDPI AG
record_format Article
series Drones
spelling doaj-art-8fdd7f00794046809b64d5eae7865b292025-08-20T03:14:39ZengMDPI AGDrones2504-446X2025-05-019538210.3390/drones9050382Adaptive Missile Avoidance Algorithm for UAV Based on Multi-Head Attention Mechanism and Dual Population Confrontation GameCheng Zhang0Junhao Song1Chengyang Tao2Zitao Su3Zhiqiang Xu4Weijia Feng5Zhaoxiang Zhang6Yuelei Xu7Unmanned System Intelligent Perception and Collaboration Technology Laboratory, Unmanned System Research Institute, Northwestern Polytechnical University, Xi’an 710129, ChinaUnmanned System Intelligent Perception and Collaboration Technology Laboratory, Unmanned System Research Institute, Northwestern Polytechnical University, Xi’an 710129, ChinaUnmanned System Intelligent Perception and Collaboration Technology Laboratory, Unmanned System Research Institute, Northwestern Polytechnical University, Xi’an 710129, ChinaUnmanned System Intelligent Perception and Collaboration Technology Laboratory, Unmanned System Research Institute, Northwestern Polytechnical University, Xi’an 710129, ChinaUnmanned System Intelligent Perception and Collaboration Technology Laboratory, Unmanned System Research Institute, Northwestern Polytechnical University, Xi’an 710129, ChinaUnmanned System Intelligent Perception and Collaboration Technology Laboratory, Unmanned System Research Institute, Northwestern Polytechnical University, Xi’an 710129, ChinaUnmanned System Intelligent Perception and Collaboration Technology Laboratory, Unmanned System Research Institute, Northwestern Polytechnical University, Xi’an 710129, ChinaUnmanned System Intelligent Perception and Collaboration Technology Laboratory, Unmanned System Research Institute, Northwestern Polytechnical University, Xi’an 710129, ChinaIn recent years, UAVs have faced increasingly severe and diversified missile threats. To address the challenge that reinforcement learning-based missile evasion algorithms struggle to adapt to various unknown missile types, we introduce a risk-sensitive PPO algorithm and propose a training framework incorporating multi-head attention mechanisms and dual-population adversarial training. The multi-head attention mechanism enables the policy network to extract latent features such as missile guidance laws from state sequences, while the dual-population adversarial approach ensures policy diversity and robustness. Compared to conventional self-play methods and GRU-based evasion strategies, our method demonstrates superior training efficiency and generates evasion policies with better adaptability to different missile types.https://www.mdpi.com/2504-446X/9/5/382missile evasionreinforcement learning intelligent control of UAVsmulti-head attention mechanismdual population confrontation game
spellingShingle Cheng Zhang
Junhao Song
Chengyang Tao
Zitao Su
Zhiqiang Xu
Weijia Feng
Zhaoxiang Zhang
Yuelei Xu
Adaptive Missile Avoidance Algorithm for UAV Based on Multi-Head Attention Mechanism and Dual Population Confrontation Game
Drones
missile evasion
reinforcement learning intelligent control of UAVs
multi-head attention mechanism
dual population confrontation game
title Adaptive Missile Avoidance Algorithm for UAV Based on Multi-Head Attention Mechanism and Dual Population Confrontation Game
title_full Adaptive Missile Avoidance Algorithm for UAV Based on Multi-Head Attention Mechanism and Dual Population Confrontation Game
title_fullStr Adaptive Missile Avoidance Algorithm for UAV Based on Multi-Head Attention Mechanism and Dual Population Confrontation Game
title_full_unstemmed Adaptive Missile Avoidance Algorithm for UAV Based on Multi-Head Attention Mechanism and Dual Population Confrontation Game
title_short Adaptive Missile Avoidance Algorithm for UAV Based on Multi-Head Attention Mechanism and Dual Population Confrontation Game
title_sort adaptive missile avoidance algorithm for uav based on multi head attention mechanism and dual population confrontation game
topic missile evasion
reinforcement learning intelligent control of UAVs
multi-head attention mechanism
dual population confrontation game
url https://www.mdpi.com/2504-446X/9/5/382
work_keys_str_mv AT chengzhang adaptivemissileavoidancealgorithmforuavbasedonmultiheadattentionmechanismanddualpopulationconfrontationgame
AT junhaosong adaptivemissileavoidancealgorithmforuavbasedonmultiheadattentionmechanismanddualpopulationconfrontationgame
AT chengyangtao adaptivemissileavoidancealgorithmforuavbasedonmultiheadattentionmechanismanddualpopulationconfrontationgame
AT zitaosu adaptivemissileavoidancealgorithmforuavbasedonmultiheadattentionmechanismanddualpopulationconfrontationgame
AT zhiqiangxu adaptivemissileavoidancealgorithmforuavbasedonmultiheadattentionmechanismanddualpopulationconfrontationgame
AT weijiafeng adaptivemissileavoidancealgorithmforuavbasedonmultiheadattentionmechanismanddualpopulationconfrontationgame
AT zhaoxiangzhang adaptivemissileavoidancealgorithmforuavbasedonmultiheadattentionmechanismanddualpopulationconfrontationgame
AT yueleixu adaptivemissileavoidancealgorithmforuavbasedonmultiheadattentionmechanismanddualpopulationconfrontationgame