Jove
Visualize
Contact Us
JoVE
x logofacebook logolinkedin logoyoutube logo
ABOUT JoVE
OverviewLeadershipBlogJoVE Help Center
AUTHORS
Publishing ProcessEditorial BoardScope & PoliciesPeer ReviewFAQSubmit
LIBRARIANS
TestimonialsSubscriptionsAccessResourcesLibrary Advisory BoardFAQ
RESEARCH
JoVE JournalMethods CollectionsJoVE Encyclopedia of ExperimentsArchive
EDUCATION
JoVE CoreJoVE BusinessJoVE Science EducationJoVE Lab ManualFaculty Resource CenterFaculty Site
Terms & Conditions of Use
Privacy Policy
Policies

Related Experiment Video

Updated: Apr 28, 2026

A Networked Desktop Virtual Reality Setup for Decision Science and Navigation Experiments with Multiple Participants
06:28

A Networked Desktop Virtual Reality Setup for Decision Science and Navigation Experiments with Multiple Participants

Published on: August 26, 2018

5.5K

Multi-Head Attention Deep Q-Network with Prioritized Experience Replay for UAV Path Planning in Dynamic Environments:

Yang Li1,2, Xinjie Qian1, Jiexin Zhang1

  • 1Department of Digital Economy and Modern Services, Yibin Polytechnic Institute of College, Yibin 644000, China.

Biomimetics (Basel, Switzerland)
|April 27, 2026
PubMed
Summary

Related Concept Videos

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by
Same author

Multi-User Computation Offloading and Resource Allocation Algorithm in a Vehicular Edge Network.

Sensors (Basel, Switzerland)·2024
Same author

A Game-Based Computing Resource Allocation Scheme of Edge Server in Vehicular Edge Computing Networks Considering Diverse Task Offloading Modes.

Sensors (Basel, Switzerland)·2024
Same author

Concentrically Encapsulated Dual-Enzyme Capsules for Synergistic Metabolic Disorder Redressing and Cytotoxic Intermediates Scavenging.

Nanomaterials (Basel, Switzerland)·2022
Same author

Investigation of Using Sky Openness Ratio as Predictor for Navigation Performance in Urban-like Environment to Support PBN in UTM.

Sensors (Basel, Switzerland)·2022
Same author

Salt crust-assisted thermal decomposition method for direct and simultaneous quantification of polypropylene microplastics and organic contaminants in high organic matter soils.

Analytica chimica acta·2022
Same author

F-box protein 17 promotes glioma progression by regulating glycolysis pathway.

Bioscience, biotechnology, and biochemistry·2022

This study introduces a novel Multi-Head Attention Deep Q-Network with Prioritized Experience Replay (MA-DQN + PER) for efficient Unmanned Aerial Vehicle (UAV) path planning. The new method significantly improves success rates, speed, and efficiency in complex environments.

Area of Science:

  • Robotics
  • Artificial Intelligence
  • Aerospace Engineering

Background:

  • Unmanned Aerial Vehicles (UAVs) are crucial for surveillance, search and rescue, and delivery.
  • Autonomous path planning for UAVs in dynamic environments with obstacles, wind, and energy constraints is challenging.

Purpose of the Study:

  • To develop an efficient UAV path planning method using deep reinforcement learning and bio-inspired attention.
  • To address challenges in dynamic environments, including moving obstacles, wind, and energy limitations.

Main Methods:

  • Proposed a novel Multi-Head Attention Deep Q-Network with Prioritized Experience Replay (MA-DQN + PER).
  • Utilized a 46-dimensional state space capturing environmental data (obstacles, wind, energy).
  • Incorporated four attention heads for selective environmental focus and employed a curriculum learning strategy with 10 difficulty levels.
Keywords:
UAV path planningbio-inspired computingcurriculum learningdeep reinforcement learningmulti-head attention

Related Experiment Videos

Last Updated: Apr 28, 2026

A Networked Desktop Virtual Reality Setup for Decision Science and Navigation Experiments with Multiple Participants
06:28

A Networked Desktop Virtual Reality Setup for Decision Science and Navigation Experiments with Multiple Participants

Published on: August 26, 2018

5.5K

Main Results:

  • Achieved a 96% task success rate, ensuring UAVs reach targets without collision or battery depletion.
  • Demonstrated a 68% improvement in convergence speed compared to baseline Deep Q-Network (DQN).
  • Showcased superior performance in path efficiency (+17%), energy reduction (-26%), and collision avoidance.

Conclusions:

  • The MA-DQN + PER approach offers a highly effective solution for autonomous UAV path planning in complex, dynamic environments.
  • The integration of attention mechanisms and prioritized experience replay significantly enhances UAV navigation capabilities.
  • This method provides a robust framework for improving UAV mission success, efficiency, and safety.