To read this content please select one of the options below:

Vision and laser fused SLAM in indoor environments with multi-robot system

Haoyao Chen (School of Mechanical Engineering and Automation, Harbin Institute of Technology Shenzen, Shenzen, China)
Hailin Huang (School of Mechanical Engineering and Automation, Harbin Institute of Technology Shenzen, Shenzen, China)
Ye Qin (School of Mechanical Engineering and Automation, Harbin Institute of Technology Shenzen, Shenzen, China)
Yanjie Li (School of Mechanical Engineering and Automation, Harbin Institute of Technology Shenzen, Shenzen, China)
Yunhui Liu (Department of Mechanical and Automation Engineering, Chinese University of Hong Kong, China)

Assembly Automation

ISSN: 0144-5154

Article publication date: 1 May 2019

Issue publication date: 3 June 2019

974

Abstract

Purpose

Multi-robot laser-based simultaneous localization and mapping (SLAM) in large-scale environments is an essential but challenging issue in mobile robotics, especially in situations wherein no prior knowledge is available between robots. Moreover, the cumulative errors of every individual robot exert a serious negative effect on loop detection and map fusion. To address these problems, this paper aims to propose an efficient approach that combines laser and vision measurements.

Design/methodology/approach

A multi-robot visual laser-SLAM is developed to realize robust and efficient SLAM in large-scale environments; both vision and laser loop detections are integrated to detect robust loops. A method based on oriented brief (ORB) feature detection and bag of words (BoW) is developed, to ensure the robustness and computational effectiveness of the multi-robot SLAM system. A robust and efficient graph fusion algorithm is proposed to merge pose graphs from different robots.

Findings

The proposed method can detect loops more quickly and accurately than the laser-only SLAM, and it can fuse the submaps of each single robot to promote the efficiency, accuracy and robustness of the system.

Originality/value

Compared with the state of art of multi-robot SLAM approaches, the paper proposed a novel and more sophisticated approach. The vision-based and laser-based loops are integrated to realize a robust loop detection. The ORB features and BoW technologies are further utilized to gain real-time performance. Finally, random sample consensus and least-square methodologies are used to remove the outlier loops among robots.

Keywords

Acknowledgements

Funding: National Natural Science Foundation of China – No. 61673131 and No. U1713206; Shenzhen Economic, Trade and Information Commission No. 20170505160946600; Shenzhen Science and Innovation Committee No. JCYJ20160427183958817; Hong Kong Research Grant Council No. 14204814; and National Natural Science Foundation of China – No. U1613218.

Citation

Chen, H., Huang, H., Qin, Y., Li, Y. and Liu, Y. (2019), "Vision and laser fused SLAM in indoor environments with multi-robot system", Assembly Automation, Vol. 39 No. 2, pp. 297-307. https://doi.org/10.1108/AA-04-2018-065

Publisher

:

Emerald Publishing Limited

Copyright © 2019, Emerald Publishing Limited

Related articles