A loop closure detection method based on semantic segmentation and convolutional neural network

• Published in: 2021 International Conference on Artificial Intelligence and Electromechanical Automation (AIEA) pp. 269 - 272
• Main Authors: Li, Hongyang, Tian, Chengjun, Wang, Lequan, Lv, Hongfu
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01.05.2021
• Subjects: closed loop detection; Convolutional Neural Network (CNN); Feature extraction; Hardware; Image segmentation; Neural networks; semantic segmentation; Semantics; Simultaneous localization and mapping; Simultaneous Localization and Mapping (SLAM); Visualization
• DOI: 10.1109/AIEA53260.2021.00063

As artificial intelligence flourishes and many related technologies continue to develop, Visual Simultaneous Localization and Mapping, as the "vision" of robots, can utilize a large amount of environmental information. In addition, semantic segmentation can distinguish the background in the image from the moving objects. In recent years, convolutional neural networks have been successfully applied to image processing, and subsequently entered the field of V-SLAM research. Related researchers have tried to directly use convolutional neural networks to extract image features for loop closure detection, but the effect has not surpassed traditional methods. In order to make full use of image information, this paper proposes a loop closure detection and SLAM method based on semantic segmentation and convolutional neural network.