A Vision-Based End-to-End Reinforcement Learning Framework for Drone Target Tracking

Drone target tracking, which involves instructing drone movement to follow a moving target, encounters several challenges: (1) traditional methods need accurate state estimation of both the drone and target; (2) conventional Proportional–Derivative (PD) controllers require tedious parameter tuning a...

Full description

Saved in:
Bibliographic Details
Main Authors: Xun Zhao, Xinjian Huang, Jianheng Cheng, Zhendong Xia, Zhiheng Tu
Format: Article
Language:English
Published: MDPI AG 2024-10-01
Series:Drones
Subjects:
Online Access:https://www.mdpi.com/2504-446X/8/11/628
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1846153745066360832
author Xun Zhao
Xinjian Huang
Jianheng Cheng
Zhendong Xia
Zhiheng Tu
author_facet Xun Zhao
Xinjian Huang
Jianheng Cheng
Zhendong Xia
Zhiheng Tu
author_sort Xun Zhao
collection DOAJ
description Drone target tracking, which involves instructing drone movement to follow a moving target, encounters several challenges: (1) traditional methods need accurate state estimation of both the drone and target; (2) conventional Proportional–Derivative (PD) controllers require tedious parameter tuning and struggle with nonlinear properties; and (3) reinforcement learning methods, though promising, rely on the drone’s self-state estimation, adding complexity and computational load and reducing reliability. To address these challenges, this study proposes an innovative model-free end-to-end reinforcement learning framework, the VTD3 (Vision-Based Twin Delayed Deep Deterministic Policy Gradient), for drone target tracking tasks. This framework focuses on controlling the drone to follow a moving target while maintaining a specific distance. VTD3 is a pure vision-based tracking algorithm which integrates the YOLOv8 detector, the BoT-SORT tracking algorithm, and the Twin Delayed Deep Deterministic Policy Gradient (TD3) algorithm. It diminishes reliance on GPS and other sensors while simultaneously enhancing the tracking capability for complex target motion trajectories. In a simulated environment, we assess the tracking performance of VTD3 across four complex target motion trajectories (triangular, square, sawtooth, and square wave, including scenarios with occlusions). The experimental results indicate that our proposed VTD3 reinforcement learning algorithm substantially outperforms conventional PD controllers in drone target tracking applications. Across various target trajectories, the VTD3 algorithm demonstrates a significant reduction in average tracking errors along the X-axis and Y-axis of up to <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mrow><mn>34.35</mn><mo>%</mo></mrow></semantics></math></inline-formula> and <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mrow><mn>45.36</mn><mo>%</mo></mrow></semantics></math></inline-formula>, respectively. Additionally, it achieves a notable improvement of up to <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mrow><mn>66.10</mn><mo>%</mo></mrow></semantics></math></inline-formula> in altitude control precision. In terms of motion smoothness, the VTD3 algorithm markedly enhances performance metrics, with improvements of up to <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mrow><mn>37.70</mn><mo>%</mo></mrow></semantics></math></inline-formula> in jitter and <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mrow><mn>60.64</mn><mo>%</mo></mrow></semantics></math></inline-formula> in Jerk RMS. Empirical results verify the superiority and feasibility of our proposed VTD3 framework for drone target tracking.
format Article
id doaj-art-ab1f69a6e7364524bc1b6a25a0456cef
institution Kabale University
issn 2504-446X
language English
publishDate 2024-10-01
publisher MDPI AG
record_format Article
series Drones
spelling doaj-art-ab1f69a6e7364524bc1b6a25a0456cef2024-11-26T18:00:37ZengMDPI AGDrones2504-446X2024-10-0181162810.3390/drones8110628A Vision-Based End-to-End Reinforcement Learning Framework for Drone Target TrackingXun Zhao0Xinjian Huang1Jianheng Cheng2Zhendong Xia3Zhiheng Tu4School of Mechanical Engineering, Nanjing University of Science and Technology, Nanjing 210094, ChinaSchool of Cyber Science and Engineering, Nanjing University of Science and Technology, Nanjing 210094, ChinaSchool of Sino-French Engineers, Nanjing University of Science and Technology, Nanjing 210094, ChinaSchool of Sino-French Engineers, Nanjing University of Science and Technology, Nanjing 210094, ChinaSchool of Mechanical Engineering, Nanjing University of Science and Technology, Nanjing 210094, ChinaDrone target tracking, which involves instructing drone movement to follow a moving target, encounters several challenges: (1) traditional methods need accurate state estimation of both the drone and target; (2) conventional Proportional–Derivative (PD) controllers require tedious parameter tuning and struggle with nonlinear properties; and (3) reinforcement learning methods, though promising, rely on the drone’s self-state estimation, adding complexity and computational load and reducing reliability. To address these challenges, this study proposes an innovative model-free end-to-end reinforcement learning framework, the VTD3 (Vision-Based Twin Delayed Deep Deterministic Policy Gradient), for drone target tracking tasks. This framework focuses on controlling the drone to follow a moving target while maintaining a specific distance. VTD3 is a pure vision-based tracking algorithm which integrates the YOLOv8 detector, the BoT-SORT tracking algorithm, and the Twin Delayed Deep Deterministic Policy Gradient (TD3) algorithm. It diminishes reliance on GPS and other sensors while simultaneously enhancing the tracking capability for complex target motion trajectories. In a simulated environment, we assess the tracking performance of VTD3 across four complex target motion trajectories (triangular, square, sawtooth, and square wave, including scenarios with occlusions). The experimental results indicate that our proposed VTD3 reinforcement learning algorithm substantially outperforms conventional PD controllers in drone target tracking applications. Across various target trajectories, the VTD3 algorithm demonstrates a significant reduction in average tracking errors along the X-axis and Y-axis of up to <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mrow><mn>34.35</mn><mo>%</mo></mrow></semantics></math></inline-formula> and <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mrow><mn>45.36</mn><mo>%</mo></mrow></semantics></math></inline-formula>, respectively. Additionally, it achieves a notable improvement of up to <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mrow><mn>66.10</mn><mo>%</mo></mrow></semantics></math></inline-formula> in altitude control precision. In terms of motion smoothness, the VTD3 algorithm markedly enhances performance metrics, with improvements of up to <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mrow><mn>37.70</mn><mo>%</mo></mrow></semantics></math></inline-formula> in jitter and <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mrow><mn>60.64</mn><mo>%</mo></mrow></semantics></math></inline-formula> in Jerk RMS. Empirical results verify the superiority and feasibility of our proposed VTD3 framework for drone target tracking.https://www.mdpi.com/2504-446X/8/11/628drone target trackingend to endreinforcement learningYOLOv8 detectorBoT-SORTtwin delayed deep deterministic policy gradient
spellingShingle Xun Zhao
Xinjian Huang
Jianheng Cheng
Zhendong Xia
Zhiheng Tu
A Vision-Based End-to-End Reinforcement Learning Framework for Drone Target Tracking
Drones
drone target tracking
end to end
reinforcement learning
YOLOv8 detector
BoT-SORT
twin delayed deep deterministic policy gradient
title A Vision-Based End-to-End Reinforcement Learning Framework for Drone Target Tracking
title_full A Vision-Based End-to-End Reinforcement Learning Framework for Drone Target Tracking
title_fullStr A Vision-Based End-to-End Reinforcement Learning Framework for Drone Target Tracking
title_full_unstemmed A Vision-Based End-to-End Reinforcement Learning Framework for Drone Target Tracking
title_short A Vision-Based End-to-End Reinforcement Learning Framework for Drone Target Tracking
title_sort vision based end to end reinforcement learning framework for drone target tracking
topic drone target tracking
end to end
reinforcement learning
YOLOv8 detector
BoT-SORT
twin delayed deep deterministic policy gradient
url https://www.mdpi.com/2504-446X/8/11/628
work_keys_str_mv AT xunzhao avisionbasedendtoendreinforcementlearningframeworkfordronetargettracking
AT xinjianhuang avisionbasedendtoendreinforcementlearningframeworkfordronetargettracking
AT jianhengcheng avisionbasedendtoendreinforcementlearningframeworkfordronetargettracking
AT zhendongxia avisionbasedendtoendreinforcementlearningframeworkfordronetargettracking
AT zhihengtu avisionbasedendtoendreinforcementlearningframeworkfordronetargettracking
AT xunzhao visionbasedendtoendreinforcementlearningframeworkfordronetargettracking
AT xinjianhuang visionbasedendtoendreinforcementlearningframeworkfordronetargettracking
AT jianhengcheng visionbasedendtoendreinforcementlearningframeworkfordronetargettracking
AT zhendongxia visionbasedendtoendreinforcementlearningframeworkfordronetargettracking
AT zhihengtu visionbasedendtoendreinforcementlearningframeworkfordronetargettracking