Loop closure detection is a critical yet still open technique to enhance the performance of Simultaneous Localization and Mapping (SLAM). In this paper, a very deep and lightweight neural network DeLightLCD is proposed to enable efficient loop closure detection. The architecture of the network contains two key modules: a siamese feature extraction module and 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 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 loop closure detection in large-scale environments. This approach was evaluated on the open-source KITTI odometry benchmark and Ford campus dataset. 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.

After the fifth wave of the COVID-19 outbreak in May 2022, the Hong Kong government decided to ease the restrictions policy step by step. The main change was to re-open some venues that people like to visit and extend the hours of operation. With the implementation of the relaxed policy, however, the number of confirmed cases rose again. As a result, further relaxation was delayed. As an evaluation of the effectiveness of the restrictions policy could be a reference for future policies balancing viral spread and functionality of society, this paper aimed to respond to this question from the spatial point distribution view. The time, from late March 2020 to February 2021, during which the related policies took place was divided into six periods based on the policy trend (tightening or relaxing). The two-variable Ripley’s Kfunction was applied for each period to explore the spatial dependence between confirmed cases and venues as changes in the spatial pattern can reveal the effect of the policy. The results show that, as time passed, the clustering degree decreased and reached its lowest level from August to mid-November 2020, then significantly increased, with the extent of clustering becoming more remarkable and the significant cluster size widening. Our results indicate that the policy had a positive effect on suppressing the spread of the virus in mid-July 2020. Then, with the virus infiltrating the community, the policy had little impact on containing the virus but likely contributed to avoid further infection.

This paper propose UAO, an acoustic device-based pedestrian tracking technique that leverages acoustic sensing to detect steps, extract step features and estimate step length, thereby calculating the walking distance. UAO enables handheld devices equipped with commercial speakers and microphones to track pedestrians without the need for calibration to achieve accurate stride length estimation (SLE) on acoustic devices. UAO performs active acoustic sensing, packaging the cross-correlation results of acoustic data with the transmitted signal template into acoustic images. Innovatively, this paper employs an acoustic image target extraction method based on few-shot learning to accurately capture step features and detect steps, further extracting detailed step parameters, including step frequency and the vertical displacement of the body's center of mass, for an adaptive SLE model. We achieve 99.1$\%$ step recognition accuracy and less than 2$\%$ SLE error across modes, including Reading, Picturing, Calling, and Swinging. Compared to existing learning-based Inertial Odometry, the proposed UAO requires only 10$\%$ of the dataset size during the training phase.