EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Improved Robot Path Planning Method Based on Deep Reinforcement Learning

Huiyan Han, Jiaqi Wang, Liqun Kuang, Xie Han, Hongxin Xue

Sensors · 2023

DOI: 10.3390/s23125622 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Sensors, 2023. Verfasst von Huiyan Han und 4 weiteren Autor:innen. Beteiligte Einrichtung: School of Computer Science and Technology, North University of China, Taiyuan 030051, China. Hauptthemen laut MEDLINE: Robotics. Die Arbeit wurde bisher 34-mal zitiert (Crossref) und nennt 28 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: National Natural Science Foundation of China, Shanxi Province Science and Technology Major Special Project, Shanxi Province Science and Technology Achievements Transformation Guidance Special Project.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

With the advancement of robotics, the field of path planning is currently experiencing a period of prosperity. Researchers strive to address this nonlinear problem and have achieved remarkable results through the implementation of the Deep Reinforcement Learning (DRL) algorithm DQN (Deep Q-Network). However, persistent challenges remain, including the curse of dimensionality, difficulties of model convergence and sparsity in rewards. To tackle these problems, this paper proposes an enhanced DDQN (Double DQN) path planning approach, in which the information after dimensionality reduction is fed into a two-branch network that incorporates expert knowledge and an optimized reward function to guide the training process. The data generated during the training phase are initially discretized into corresponding low-dimensional spaces. An “expert experience” module is introduced to facilitate the model’s early-stage training acceleration in the Epsilon–Greedy algorithm. To tackle navigation and obstacle avoidance separately, a dual-branch network structure is presented. We further optimize the reward function enabling intelligent agents to receive prompt feedback from the environment after performing each action. Experiments conducted in both virtual and real-world environments have demonstrated that the enhanced algorithm can accelerate model convergence, improve training stability and generate a smooth, shorter and collision-free path.

Schlagwörter

MeSH-Schlagwörter (MEDLINE)

RoboticsRewardAccelerationAlgorithmsIntelligence

Stichwörter der Autor:innen

DDQNdeep reinforcement learningexpert experiencerobot path planning

Autor:innen und Institutionen

  • School of Computer Science and Technology, North University of China, Taiyuan 030051, China.
  • Shanxi Key Laboratory of Machine Vision and Virtual Reality, Taiyuan 030051, China.
  • Shanxi Vision Information Processing and Intelligent Robot Engineering Research Center, Taiyuan 030051, China.

Förderung

  • National Natural Science Foundation of China
  • Shanxi Province Science and Technology Major Special Project
  • Shanxi Province Science and Technology Achievements Transformation Guidance Special Project
  • Shanxi Key Laboratory of Machine Vision and Virtual Reality

Zitieren

APA

Han, H., Wang, J., Kuang, L., Han, X., & Xue, H. (2023). Improved Robot Path Planning Method Based on Deep Reinforcement Learning. Sensors. https://doi.org/10.3390/s23125622

Vancouver

Han H, Wang J, Kuang L, Han X, Xue H. Improved Robot Path Planning Method Based on Deep Reinforcement Learning. Sensors. 2023. doi:10.3390/s23125622

BibTeX
@article{Han2023,
  author = {Han, Huiyan and Wang, Jiaqi and Kuang, Liqun and Han, Xie and Xue, Hongxin},
  title = {Improved Robot Path Planning Method Based on Deep Reinforcement Learning},
  journal = {Sensors},
  year = {2023},
  doi = {10.3390/s23125622}
}
Kostenlos starten →