Improved duelling deep Q-networks based path planning for intelligent agents

The natural deep Q-network (DQN) usually requires a long training time because the data usage efficiency is relatively low due to uniform sampling. Importance sampling (IS) can promote important experiences and is more efficient in the neural network training process. In this paper, an efficient learning mechanism using the IS technique is incorporated into duelling DQN algorithm, and is further applied to path planning task for an agent. Different from the traditional DQN algorithm, proposed algorithm improves the sampling efficiency. In this experiment, four target points on the map are deployed to evaluate the loss and the accumulated reward. Simulations and comparisons in various simulation situations demonstrate the effectiveness and superiority of the proposed path planning scheme for an intelligent agent.

• Record URL:
  https://doi.org/10.1504/IJVD.2023.131056
• Availability:
  • Find a library where document is available. Order URL: http://worldcat.org/issn/14775360
• Supplemental Notes:
  • Copyright © 2023 Inderscience Enterprises Ltd.
• Authors:
  • Lin, Yejin
  • Wen, Jiayi
• Publication Date: 2023

Language

• English

Media Info

• Media Type: Web
• Features: References;
• Pagination: pp 232-247
• Serial:
  • International Journal of Vehicle Design
  • Volume: 91
  • Issue Number: 1-3
  • Publisher: Inderscience Enterprises Limited
  • ISSN: 1477-5360
  • Serial URL: http://www.inderscience.com/jhome.php?jcode=IJVD

Subject/Index Terms

• TRT Terms: Intelligent agents; Machine learning; Neural networks; Sampling; Trajectory control
• Subject Areas: Data and Information Technology; Highways; Vehicles and Equipment;

Filing Info

• Accession Number: 01900117
• Record Type: Publication
• Files: TRIS
• Created Date: Nov 20 2023 9:12AM