Path planning for active SLAM based on deep reinforcement learning under unknown environments

Shuhuan Wen (Lead / Corresponding author), Yanfang Zhao, Xiao Yuan, Zongtao Wang, Dan Zhang, Luigi Manfredi

Research output: Contribution to journalArticle

Abstract

Autonomous navigation in complex environment is an important requirement for the design of a robot. Active SLAM (simultaneous localization and mapping) combining, which combine path planning with SLAM, is proposed to improve the ability of autonomous navigation in complex environment. In this paper, fully convolutional residual networks are used to recognize the obstacles to get depth image. The avoidance obstacle path is planned by Dueling DQN algorithm in the robot’s navigation; at the same time, the 2D map of the environment is built based on FastSLAM. The experiments show that the proposed algorithm can successfully identify and avoid moving and static obstacles with different quantities in the environment, and realize the autonomous navigation of the robot in a complex environment.

Original languageEnglish
Number of pages10
JournalIntelligent Service Robotics
Early online date16 Jan 2020
DOIs
Publication statusE-pub ahead of print - 16 Jan 2020

Fingerprint

Reinforcement learning
Motion planning
Navigation
Robots
Collision avoidance
Experiments

Keywords

  • Path planning
  • FastSLAM
  • Deep reinforcement learning

Cite this

Wen, Shuhuan ; Zhao, Yanfang ; Yuan, Xiao ; Wang, Zongtao ; Zhang, Dan ; Manfredi, Luigi. / Path planning for active SLAM based on deep reinforcement learning under unknown environments. In: Intelligent Service Robotics. 2020.
@article{4f412c40446a4c26a47db694aa796870,
title = "Path planning for active SLAM based on deep reinforcement learning under unknown environments",
abstract = "Autonomous navigation in complex environment is an important requirement for the design of a robot. Active SLAM (simultaneous localization and mapping) combining, which combine path planning with SLAM, is proposed to improve the ability of autonomous navigation in complex environment. In this paper, fully convolutional residual networks are used to recognize the obstacles to get depth image. The avoidance obstacle path is planned by Dueling DQN algorithm in the robot’s navigation; at the same time, the 2D map of the environment is built based on FastSLAM. The experiments show that the proposed algorithm can successfully identify and avoid moving and static obstacles with different quantities in the environment, and realize the autonomous navigation of the robot in a complex environment.",
keywords = "Path planning, FastSLAM, Deep reinforcement learning",
author = "Shuhuan Wen and Yanfang Zhao and Xiao Yuan and Zongtao Wang and Dan Zhang and Luigi Manfredi",
note = "This work was supported by the National Natural Science Foundation of China under Grant No. 61773333.",
year = "2020",
month = "1",
day = "16",
doi = "10.1007/s11370-019-00310-w",
language = "English",

}

Path planning for active SLAM based on deep reinforcement learning under unknown environments. / Wen, Shuhuan (Lead / Corresponding author); Zhao, Yanfang; Yuan, Xiao; Wang, Zongtao; Zhang, Dan; Manfredi, Luigi.

In: Intelligent Service Robotics, 16.01.2020.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Path planning for active SLAM based on deep reinforcement learning under unknown environments

AU - Wen, Shuhuan

AU - Zhao, Yanfang

AU - Yuan, Xiao

AU - Wang, Zongtao

AU - Zhang, Dan

AU - Manfredi, Luigi

N1 - This work was supported by the National Natural Science Foundation of China under Grant No. 61773333.

PY - 2020/1/16

Y1 - 2020/1/16

N2 - Autonomous navigation in complex environment is an important requirement for the design of a robot. Active SLAM (simultaneous localization and mapping) combining, which combine path planning with SLAM, is proposed to improve the ability of autonomous navigation in complex environment. In this paper, fully convolutional residual networks are used to recognize the obstacles to get depth image. The avoidance obstacle path is planned by Dueling DQN algorithm in the robot’s navigation; at the same time, the 2D map of the environment is built based on FastSLAM. The experiments show that the proposed algorithm can successfully identify and avoid moving and static obstacles with different quantities in the environment, and realize the autonomous navigation of the robot in a complex environment.

AB - Autonomous navigation in complex environment is an important requirement for the design of a robot. Active SLAM (simultaneous localization and mapping) combining, which combine path planning with SLAM, is proposed to improve the ability of autonomous navigation in complex environment. In this paper, fully convolutional residual networks are used to recognize the obstacles to get depth image. The avoidance obstacle path is planned by Dueling DQN algorithm in the robot’s navigation; at the same time, the 2D map of the environment is built based on FastSLAM. The experiments show that the proposed algorithm can successfully identify and avoid moving and static obstacles with different quantities in the environment, and realize the autonomous navigation of the robot in a complex environment.

KW - Path planning

KW - FastSLAM

KW - Deep reinforcement learning

UR - http://www.scopus.com/inward/record.url?scp=85078619753&partnerID=8YFLogxK

U2 - 10.1007/s11370-019-00310-w

DO - 10.1007/s11370-019-00310-w

M3 - Article

ER -