Cargando…

Non-Communication Decentralized Multi-Robot Collision Avoidance in Grid Map Workspace with Double Deep Q-Network

This paper presents a novel decentralized multi-robot collision avoidance method with deep reinforcement learning, which is not only suitable for the large-scale grid map workspace multi-robot system, but also directly processes Lidar signals instead of communicating between the robots. According to...

Descripción completa

Detalles Bibliográficos
Autores principales: Chen, Lin, Zhao, Yongting, Zhao, Huanjun, Zheng, Bin
Formato: Online Artículo Texto
Lenguaje:English
Publicado: MDPI 2021
Materias:
Acceso en línea:https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7866139/
https://www.ncbi.nlm.nih.gov/pubmed/33513856
http://dx.doi.org/10.3390/s21030841
_version_ 1783648011458445312
author Chen, Lin
Zhao, Yongting
Zhao, Huanjun
Zheng, Bin
author_facet Chen, Lin
Zhao, Yongting
Zhao, Huanjun
Zheng, Bin
author_sort Chen, Lin
collection PubMed
description This paper presents a novel decentralized multi-robot collision avoidance method with deep reinforcement learning, which is not only suitable for the large-scale grid map workspace multi-robot system, but also directly processes Lidar signals instead of communicating between the robots. According to the particularity of the workspace, we handcrafted a reward function, which considers both the collision avoidance among the robots and as little as possible change of direction of the robots during driving. Using Double Deep Q-Network (DDQN), the policy was trained in the simulation grid map workspace. By designing experiments, we demonstrated that the learned policy can guide the robot well to effectively travel from the initial position to the goal position in the grid map workspace and to avoid collisions with others while driving.
format Online
Article
Text
id pubmed-7866139
institution National Center for Biotechnology Information
language English
publishDate 2021
publisher MDPI
record_format MEDLINE/PubMed
spelling pubmed-78661392021-02-07 Non-Communication Decentralized Multi-Robot Collision Avoidance in Grid Map Workspace with Double Deep Q-Network Chen, Lin Zhao, Yongting Zhao, Huanjun Zheng, Bin Sensors (Basel) Article This paper presents a novel decentralized multi-robot collision avoidance method with deep reinforcement learning, which is not only suitable for the large-scale grid map workspace multi-robot system, but also directly processes Lidar signals instead of communicating between the robots. According to the particularity of the workspace, we handcrafted a reward function, which considers both the collision avoidance among the robots and as little as possible change of direction of the robots during driving. Using Double Deep Q-Network (DDQN), the policy was trained in the simulation grid map workspace. By designing experiments, we demonstrated that the learned policy can guide the robot well to effectively travel from the initial position to the goal position in the grid map workspace and to avoid collisions with others while driving. MDPI 2021-01-27 /pmc/articles/PMC7866139/ /pubmed/33513856 http://dx.doi.org/10.3390/s21030841 Text en © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).
spellingShingle Article
Chen, Lin
Zhao, Yongting
Zhao, Huanjun
Zheng, Bin
Non-Communication Decentralized Multi-Robot Collision Avoidance in Grid Map Workspace with Double Deep Q-Network
title Non-Communication Decentralized Multi-Robot Collision Avoidance in Grid Map Workspace with Double Deep Q-Network
title_full Non-Communication Decentralized Multi-Robot Collision Avoidance in Grid Map Workspace with Double Deep Q-Network
title_fullStr Non-Communication Decentralized Multi-Robot Collision Avoidance in Grid Map Workspace with Double Deep Q-Network
title_full_unstemmed Non-Communication Decentralized Multi-Robot Collision Avoidance in Grid Map Workspace with Double Deep Q-Network
title_short Non-Communication Decentralized Multi-Robot Collision Avoidance in Grid Map Workspace with Double Deep Q-Network
title_sort non-communication decentralized multi-robot collision avoidance in grid map workspace with double deep q-network
topic Article
url https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7866139/
https://www.ncbi.nlm.nih.gov/pubmed/33513856
http://dx.doi.org/10.3390/s21030841
work_keys_str_mv AT chenlin noncommunicationdecentralizedmultirobotcollisionavoidanceingridmapworkspacewithdoubledeepqnetwork
AT zhaoyongting noncommunicationdecentralizedmultirobotcollisionavoidanceingridmapworkspacewithdoubledeepqnetwork
AT zhaohuanjun noncommunicationdecentralizedmultirobotcollisionavoidanceingridmapworkspacewithdoubledeepqnetwork
AT zhengbin noncommunicationdecentralizedmultirobotcollisionavoidanceingridmapworkspacewithdoubledeepqnetwork