期刊
IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS
卷 68, 期 4, 页码 3577-3587出版社
IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC
DOI: 10.1109/TIE.2020.2982096
关键词
Simultaneous localization and mapping; Visualization; Training; Three-dimensional displays; Optimization; Pose estimation; Depth estimation; machine learning; recurrent convolutional neural network (RCNN); simultaneous localization and mapping (SLAM); unsupervised deep learning (DL)
资金
- National Natural Science Foundation of China [61903377]
- Engineering and Physical Sciences Research Council (EPSRC) Robotics and Artificial Intelligence Offshore Robotics for Certification of Assets (ORCA) Hub [EP/R026173/1]
- EU H2020 Program under EUMarineRobots Project [731103]
- DeepField Project [857339]
- EPSRC [EP/R026173/1] Funding Source: UKRI
DeepSLAM is an unsupervised deep learning-based visual SLAM system that uses stereo imagery for training and monocular image sequences for testing. It consists of essential components such as Mapping-Net, Tracking-Net, and Loop-Net, which enable it to generate pose estimates, depth maps, and outlier rejection masks simultaneously. The system demonstrates good performance in pose estimation accuracy and robustness in challenging scenes, as evaluated on various datasets.
In this article, we propose DeepSLAM, a novel unsupervised deep learning based visual simultaneous localization and mapping (SLAM) system. The DeepSLAM training is fully unsupervised since it only requires stereo imagery instead of annotating ground-truth poses. Its testing takes a monocular image sequence as the input. Therefore, it is a monocular SLAM paradigm. DeepSLAM consists of several essential components, including Mapping-Net, Tracking-Net, Loop-Net, and a graph optimization unit. Specifically, the Mapping-Net is an encoder and decoder architecture for describing the 3-D structure of environment, whereas the Tracking-Net is a recurrent convolutional neural network architecture for capturing the camera motion. The Loop-Net is a pretrained binary classifier for detecting loop closures. DeepSLAM can simultaneously generate pose estimate, depth map, and outlier rejection mask. In this article, we evaluate its performance on various datasets, and find that DeepSLAM achieves good performance in terms of pose estimation accuracy, and is robust in some challenging scenes.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据