EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Coverage Path Planning Using Actor–Critic Deep Reinforcement Learning

Sergio Isahí Garrido-Castañeda iD, Juan Irving Vasquez iD, Mayra Antonio-Cruz iD

Sensors · 05.03.2025 · 25 (5) · 1592

DOI: 10.3390/s25051592 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Sensors, online erschienen am 05.03.2025. Verfasst von Sergio Isahí Garrido-Castañeda und 2 weiteren Autor:innen. Beteiligte Einrichtung: Centro de Innovación y Desarrollo Tecnológico en Cómputo (CIDETEC), Instituto Politécnico Nacional (IPN), Mexico City 07700, Mexico. Die Arbeit wurde bisher 14-mal zitiert (Crossref) und nennt 35 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Instituto Politécnico Nacional, Secretaría de Investigación y Posgrado (SIP), Instituto Politécnico Nacional.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

One of the main capabilities a mobile robot must demonstrate is the ability to explore its environment. The core challenge in exploration lies in planning the route to fully cover the environment. Despite recent advances, this problem remains unsolved. This study proposes an approach to address the coverage path planning problem, where the mobile robot is tasked with exploring and completely covering a terrain using a deep reinforcement learning framework. The environment is divided into cells, with obstacles designated as prohibited areas. The robot is trained using two state-of-the-art reinforcement learning algorithms based on actor–critic methods: Advantage Actor–Critic (A2C) and Proximal Policy Optimization (PPO). By defining a set of observations, states, and a reward function tailored to characteristics of the environment and the desired behavior of the robot, the training process is conducted, resulting in optimized policies for each algorithm. Then, these policies are evaluated to determine the most effective approach to accomplish the proposed task. Our findings demonstrate that actor–critic methods can produce policies capable of guiding a robot to efficiently explore and cover new environments.

Schlagwörter

Stichwörter der Autor:innen

advantage actor–criticcoverage path planningdeep reinforcement learningproximal policy optimization

Autor:innen und Institutionen

  • Centro de Innovación y Desarrollo Tecnológico en Cómputo (CIDETEC), Instituto Politécnico Nacional (IPN), Mexico City 07700, Mexico.
  • Sección de Estudios de Posgrado e Investigación, Unidad Profesional Interdisciplinaria de Ingeniería y Ciencias Sociales y Administrativas (SEPI-UPIICSA), Instituto Politécnico Nacional (IPN), Mexico City 08400, Mexico.

ORCID-Profile: Sergio Isahí Garrido-Castañeda, Juan Irving Vasquez, Mayra Antonio-Cruz

Förderung

  • Instituto Politécnico Nacional
  • Secretaría de Investigación y Posgrado (SIP), Instituto Politécnico Nacional

Zitieren

APA

Garrido-Castañeda, S. I., Vasquez, J. I., & Antonio-Cruz, M. (2025). Coverage Path Planning Using Actor–Critic Deep Reinforcement Learning. Sensors, 25(5), 1592. https://doi.org/10.3390/s25051592

Vancouver

Garrido-Castañeda SI, Vasquez JI, Antonio-Cruz M. Coverage Path Planning Using Actor–Critic Deep Reinforcement Learning. Sensors. 2025;25(5):1592. doi:10.3390/s25051592

BibTeX
@article{GarridoCastaeda2025,
  author = {Garrido-Castañeda, Sergio Isahí and Vasquez, Juan Irving and Antonio-Cruz, Mayra},
  title = {Coverage Path Planning Using Actor–Critic Deep Reinforcement Learning},
  journal = {Sensors},
  year = {2025},
  volume = {25},
  number = {5},
  pages = {1592},
  doi = {10.3390/s25051592}
}
Kostenlos starten →