Unmanned Aerial Vehicle Path Planning in Complex Dynamic Environments Based on Deep Reinforcement Learning

In this paper, an enhanced deep reinforcement learning approach is presented for unmanned aerial vehicles (UAVs) operating in dynamic and potentially hazardous environments. Initially, the capability to discern obstacles from visual data is achieved through the application of the Yolov8-StrongSort t...

Full description

Saved in:
Bibliographic Details
Main Authors: Jiandong Liu, Wei Luo, Guoqing Zhang, Ruihao Li
Format: Article
Language:English
Published: MDPI AG 2025-02-01
Series:Machines
Subjects:
Online Access:https://www.mdpi.com/2075-1702/13/2/162
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849718709700001792
author Jiandong Liu
Wei Luo
Guoqing Zhang
Ruihao Li
author_facet Jiandong Liu
Wei Luo
Guoqing Zhang
Ruihao Li
author_sort Jiandong Liu
collection DOAJ
description In this paper, an enhanced deep reinforcement learning approach is presented for unmanned aerial vehicles (UAVs) operating in dynamic and potentially hazardous environments. Initially, the capability to discern obstacles from visual data is achieved through the application of the Yolov8-StrongSort technique. Concurrently, a novel data storage system for deep Q-networks (DQN), named dynamic data memory (DDM), is introduced to hasten the learning process and convergence for UAVs. Furthermore, addressing the issue of UAVs’ paths veering too close to obstacles, a novel strategy employing an artificial potential field to adjust the reward function is introduced, which effectively guides the UAVs away from proximate obstacles. Rigorous simulation tests in an AirSim-based environment confirm the effectiveness of these methods. Compared to DQN, dueling DQN, M-DQN, improved Q-learning, DDM-DQN, EPF (enhanced potential field), APF-DQN, and L1-MBRL, our algorithm achieves the highest success rate of 77.67%, while also having the lowest average number of moving steps. Additionally, we conducted obstacle avoidance experiments with UAVs with different densities of obstacles. These tests highlight fast learning convergence and real-time obstacle detection and avoidance, ensuring successful achievement of the target.
format Article
id doaj-art-3d96d8b91d544f4f99f8244b158cece4
institution DOAJ
issn 2075-1702
language English
publishDate 2025-02-01
publisher MDPI AG
record_format Article
series Machines
spelling doaj-art-3d96d8b91d544f4f99f8244b158cece42025-08-20T03:12:19ZengMDPI AGMachines2075-17022025-02-0113216210.3390/machines13020162Unmanned Aerial Vehicle Path Planning in Complex Dynamic Environments Based on Deep Reinforcement LearningJiandong Liu0Wei Luo1Guoqing Zhang2Ruihao Li3North China Institute of Aerospace Engineering, School of Remote Sensing and Information Engineering, Langfang 065000, ChinaNorth China Institute of Aerospace Engineering, School of Remote Sensing and Information Engineering, Langfang 065000, ChinaNorth China Institute of Aerospace Engineering, School of Remote Sensing and Information Engineering, Langfang 065000, ChinaNorth China Institute of Aerospace Engineering, School of Remote Sensing and Information Engineering, Langfang 065000, ChinaIn this paper, an enhanced deep reinforcement learning approach is presented for unmanned aerial vehicles (UAVs) operating in dynamic and potentially hazardous environments. Initially, the capability to discern obstacles from visual data is achieved through the application of the Yolov8-StrongSort technique. Concurrently, a novel data storage system for deep Q-networks (DQN), named dynamic data memory (DDM), is introduced to hasten the learning process and convergence for UAVs. Furthermore, addressing the issue of UAVs’ paths veering too close to obstacles, a novel strategy employing an artificial potential field to adjust the reward function is introduced, which effectively guides the UAVs away from proximate obstacles. Rigorous simulation tests in an AirSim-based environment confirm the effectiveness of these methods. Compared to DQN, dueling DQN, M-DQN, improved Q-learning, DDM-DQN, EPF (enhanced potential field), APF-DQN, and L1-MBRL, our algorithm achieves the highest success rate of 77.67%, while also having the lowest average number of moving steps. Additionally, we conducted obstacle avoidance experiments with UAVs with different densities of obstacles. These tests highlight fast learning convergence and real-time obstacle detection and avoidance, ensuring successful achievement of the target.https://www.mdpi.com/2075-1702/13/2/162UAV obstacle avoidanceartificial potential fielddynamic environmentDQN algorithmYolov8
spellingShingle Jiandong Liu
Wei Luo
Guoqing Zhang
Ruihao Li
Unmanned Aerial Vehicle Path Planning in Complex Dynamic Environments Based on Deep Reinforcement Learning
Machines
UAV obstacle avoidance
artificial potential field
dynamic environment
DQN algorithm
Yolov8
title Unmanned Aerial Vehicle Path Planning in Complex Dynamic Environments Based on Deep Reinforcement Learning
title_full Unmanned Aerial Vehicle Path Planning in Complex Dynamic Environments Based on Deep Reinforcement Learning
title_fullStr Unmanned Aerial Vehicle Path Planning in Complex Dynamic Environments Based on Deep Reinforcement Learning
title_full_unstemmed Unmanned Aerial Vehicle Path Planning in Complex Dynamic Environments Based on Deep Reinforcement Learning
title_short Unmanned Aerial Vehicle Path Planning in Complex Dynamic Environments Based on Deep Reinforcement Learning
title_sort unmanned aerial vehicle path planning in complex dynamic environments based on deep reinforcement learning
topic UAV obstacle avoidance
artificial potential field
dynamic environment
DQN algorithm
Yolov8
url https://www.mdpi.com/2075-1702/13/2/162
work_keys_str_mv AT jiandongliu unmannedaerialvehiclepathplanningincomplexdynamicenvironmentsbasedondeepreinforcementlearning
AT weiluo unmannedaerialvehiclepathplanningincomplexdynamicenvironmentsbasedondeepreinforcementlearning
AT guoqingzhang unmannedaerialvehiclepathplanningincomplexdynamicenvironmentsbasedondeepreinforcementlearning
AT ruihaoli unmannedaerialvehiclepathplanningincomplexdynamicenvironmentsbasedondeepreinforcementlearning