Xiang, HaodongHaodongXiangZhu, XiaoshengXiaoshengZhuShi, WenzhongWenzhongShiFan, WenzhengWenzhengFanChen, PengxinPengxinChenDr. BAO Sheng2026-09-042026-09-042022IEEE Sensors Journal, 2022, vol. 22(21), pp. 20761-20772.1530-437X1558-1748http://hdl.handle.net/20.500.11861/28995Loop closure detection (LCD) is a critical yet still open technique to enhance the performance of simultaneous localization and mapping (SLAM). In this article, a very deep and lightweight neural network deep and lightweight loop closure detection (DeLightLCD) is proposed to enable efficient LCD. The architecture of the network contains two key modules: 1) a Siamese feature extraction module and 2) a dual-attention-based feature difference module. A very deep but lightweight feature extraction network is designed to extract high-dimensional and discriminative features. The dual-attention-based feature difference network generates feature difference maps to identify the difference between a pair of light detection and ranging (LiDAR) scans. Besides, a loop closure candidate rapid selection method is proposed to extract loop closure candidates from the LiDAR scan sequence. The workflow enables real-time LCD in large-scale environments. This approach was evaluated on the open-source KITTI odometry benchmark and Ford campus datasets. The experimental results show that this method outperforms the state-of-the-art methods. Although the model was trained only on the KITTI dataset, it also demonstrated superior performance on the Ford campus dataset. In particular, the proposed approach is more lightweight and highly efficient than the current existing approaches.enLight Detection And RangingSimultaneous Localization And MappingLoop ClosureLoop Closure DetectionNeural NetworkDeep Neural NetworkDiscriminative FeaturesHigh-Dimensional FeatureFeature Extraction NetworkDeep Feature ExtractionRapid SelectionKITTI DatasetLarge-scale EnvironmentsLaser ScanningLocal FeaturesPoint CloudSemantic InformationStraightforward WayDepth ImagesArea Under CurveScan PairsDual AttentionDepthwise Separable ConvolutionChannel-wise AttentionCurrent ScanRaw Point CloudRecall RateGlobal Average PoolingGlobal DescriptorsPlace RecognitionPeer Reviewed Journal Article10.1109/JSEN.2022.3206506