End-to-end deep reinforcement learning for multi-agent collaborative exploration

Exploring an unknown environment by multiple autonomous robots is a major challenge in robotics domains. As multiple robots are assigned to explore different locations, they may interfere each other making the overall tasks less efficient. In this paper, we present a new model called CNN-based Multi...

Full description

Saved in:

Bibliographic Details
Main Authors:	Chen, Zichen, Subagdja, Bhuditama, Tan, Ah-Hwee
Other Authors:	School of Electrical and Electronic Engineering
Format:	Conference or Workshop Item
Language:	English
Published:	2021
Subjects:	Engineering::Computer science and engineering::Computing methodologies::Artificial intelligence Multi-agent Exploration Deep Learning
Online Access:	https://hdl.handle.net/10356/148510
Tags:	Add Tag No Tags, Be the first to tag this record!
Institution:	Nanyang Technological University
Language:	English

id	sg-ntu-dr.10356-148510
record_format	dspace
spelling	sg-ntu-dr.10356-1485102021-05-25T09:09:51Z End-to-end deep reinforcement learning for multi-agent collaborative exploration Chen, Zichen Subagdja, Bhuditama Tan, Ah-Hwee School of Electrical and Electronic Engineering 2019 IEEE International Conference on Agents (ICA) ST Engineering-NTU Corporate Lab Engineering::Computer science and engineering::Computing methodologies::Artificial intelligence Multi-agent Exploration Deep Learning Exploring an unknown environment by multiple autonomous robots is a major challenge in robotics domains. As multiple robots are assigned to explore different locations, they may interfere each other making the overall tasks less efficient. In this paper, we present a new model called CNN-based Multi-agent Proximal Policy Optimization (CMAPPO) to multi-agent exploration wherein the agents learn the effective strategy to allocate and explore the environment using a new deep reinforcement learning architecture. The model combines convolutional neural network to process multi-channel visual inputs, curriculum-based learning, and PPO algorithm for motivation based reinforcement learning. Evaluations show that the proposed method can learn more efficient strategy for multiple agents to explore the environment than the conventional frontier-based method. National Research Foundation (NRF) Accepted version 2021-05-25T09:09:51Z 2021-05-25T09:09:51Z 2019 Conference Paper Chen, Z., Subagdja, B. & Tan, A. (2019). End-to-end deep reinforcement learning for multi-agent collaborative exploration. 2019 IEEE International Conference on Agents (ICA), 99-102. https://dx.doi.org/10.1109/AGENTS.2019.8929192 9781728140261 https://hdl.handle.net/10356/148510 10.1109/AGENTS.2019.8929192 2-s2.0-85077815398 99 102 en © 2019 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works. The published version is available at: https://doi.org/10.1109/AGENTS.2019.8929192 application/pdf
institution	Nanyang Technological University
building	NTU Library
continent	Asia
country	Singapore Singapore
content_provider	NTU Library
collection	DR-NTU
language	English
topic	Engineering::Computer science and engineering::Computing methodologies::Artificial intelligence Multi-agent Exploration Deep Learning
spellingShingle	Engineering::Computer science and engineering::Computing methodologies::Artificial intelligence Multi-agent Exploration Deep Learning Chen, Zichen Subagdja, Bhuditama Tan, Ah-Hwee End-to-end deep reinforcement learning for multi-agent collaborative exploration
description	Exploring an unknown environment by multiple autonomous robots is a major challenge in robotics domains. As multiple robots are assigned to explore different locations, they may interfere each other making the overall tasks less efficient. In this paper, we present a new model called CNN-based Multi-agent Proximal Policy Optimization (CMAPPO) to multi-agent exploration wherein the agents learn the effective strategy to allocate and explore the environment using a new deep reinforcement learning architecture. The model combines convolutional neural network to process multi-channel visual inputs, curriculum-based learning, and PPO algorithm for motivation based reinforcement learning. Evaluations show that the proposed method can learn more efficient strategy for multiple agents to explore the environment than the conventional frontier-based method.
author2	School of Electrical and Electronic Engineering
author_facet	School of Electrical and Electronic Engineering Chen, Zichen Subagdja, Bhuditama Tan, Ah-Hwee
format	Conference or Workshop Item
author	Chen, Zichen Subagdja, Bhuditama Tan, Ah-Hwee
author_sort	Chen, Zichen
title	End-to-end deep reinforcement learning for multi-agent collaborative exploration
title_short	End-to-end deep reinforcement learning for multi-agent collaborative exploration
title_full	End-to-end deep reinforcement learning for multi-agent collaborative exploration
title_fullStr	End-to-end deep reinforcement learning for multi-agent collaborative exploration
title_full_unstemmed	End-to-end deep reinforcement learning for multi-agent collaborative exploration
title_sort	end-to-end deep reinforcement learning for multi-agent collaborative exploration
publishDate	2021
url	https://hdl.handle.net/10356/148510
_version_	1701270615776821248

End-to-end deep reinforcement learning for multi-agent collaborative exploration

Similar Items