Article
Engineering, Electrical & Electronic
Ioannis Tsampikos Papapetros, Ioannis Kansizoglou, Loukas Bampis, Antonios Gasteratos
Summary: In this article, a sequence-based technique is proposed to remap sequential representation data into the distance-space domain and normalize them, effectively eliminating the negative impact of environmental changes on visual place recognition. The framework shows significant performance improvement over other approaches.
Article
Engineering, Electrical & Electronic
Chunbo Mei, Zhenhui Fan, Qiju Zhu, Pengxiang Yang, Zhenhuan Hou, Honglong Jin
Summary: This article aims to develop an accurate, fast, and robust scene matching navigation system (SMNS) based on vision/inertial fusion for unmanned aerial vehicles (UAVs). It proposes a method to improve the georeferencing of aerial images and reduce projection errors by optimizing the homography matrix. A robust noise processing strategy and an improved feature extraction algorithm are introduced to eliminate variations due to climate, time, and season, ensuring accuracy in the matching procedure. A novel matching strategy based on logic graphs is designed under the framework of the SMNS, which enhances the efficiency and effectiveness of the system. Experimental results demonstrate that the proposed SMNS outperforms existing strategies in terms of matching aerial and satellite images.
IEEE SENSORS JOURNAL
(2023)
Article
Robotics
Maria Waheed, Michael Milford, Klaus McDonald-Maier, Shoaib Ehsan
Summary: This paper investigates the complementarity of state-of-the-art VPR methods and proposes a framework to assess their combination effects, showing the potential of different combinations of techniques for achieving better performance.
IEEE ROBOTICS AND AUTOMATION LETTERS
(2021)
Article
Automation & Control Systems
Saba Arshad, Gon-Woo Kim
Summary: This research focuses on the selection and matching of robust features for accurate place recognition in challenging environments, and highlights the limitations of existing research. A fast and effective visual place recognition method is proposed that integrates the Bag-of-Words (BoW) vocabulary with a robust feature matcher. The method reduces the search space for image matching and enhances place matching performance by identifying and removing inconsistent feature matches. The proposed method is evaluated on benchmark datasets and compared with state-of-the-art VPR methods, demonstrating its effectiveness.
INTERNATIONAL JOURNAL OF CONTROL AUTOMATION AND SYSTEMS
(2023)
Article
Computer Science, Information Systems
Dong Fu, Hao Xia, Yujie Liu, Yanyou Qiao
Summary: This study proposes a monocular visual-inertial navigation system, VINS-dimc, which improves the positioning accuracy of VINS in dynamic environments by integrating various constraints. The system effectively eliminates dynamic feature points while preserving static ones, enhancing the robustness and accuracy of VINS.
ISPRS INTERNATIONAL JOURNAL OF GEO-INFORMATION
(2022)
Article
Multidisciplinary Sciences
Saed Alqaraleh, A. H. Abdul Hafez, Ammar Tello
Summary: This paper introduces a new visual place recognition method based on dynamic time warping and deep convolutional neural network, which is capable of addressing changes in visual conditions like appearance and viewpoint. Experimental results demonstrate that the proposed method outperforms traditional features and is compared with other existing visual place recognition algorithms.
ARABIAN JOURNAL FOR SCIENCE AND ENGINEERING
(2021)
Article
Engineering, Electrical & Electronic
Chen Fan, Zhouwen Zhou, Xiaofeng He, Ying Fan, Lilian Zhang, Xuesong Wu, Xiaoping Hu
Summary: This paper presents a bio-inspired multi-sensor navigation system for UAVs in GNSS-denied environments. The system integrates outputs from a skylight polarized sensor, a micro-inertia sensor, and a monocular camera to provide reliable position and heading constraints. An optimal orientation algorithm and a two-dimensional visual place recognition are proposed to improve navigation performance. Experimental results demonstrate that the proposed system outperforms other vision-based navigation algorithms in terms of position accuracy.
IEEE SENSORS JOURNAL
(2022)
Article
Robotics
Hanjing Ye, Weinan Chen, Jingwen Yu, Li He, Yisheng Guan, Hong Zhang
Summary: This paper proposes a method that uses a convolutional autoencoder (CAE) to solve the problem of visual place recognition (VPR) in condition-varying environments. The authors utilize a high-level layer of a pre-trained convolutional neural network (CNN) to generate features and train a CAE to map these features to a low-dimensional space, improving the condition invariance property of the descriptor and reducing its dimension. The method is verified on four challenging real-world datasets with significant illumination changes, showing superiority to the state-of-the-art. The code for this work is publicly available at https://github.com/MedlarTea/CAE-VPR.
Article
Engineering, Multidisciplinary
Peng Ding, Xianghong Cheng
Summary: Terrain-aided strapdown inertial navigation system is a positioning technique for underwater vehicles. This study proposes a two-stage combined matching algorithm to overcome the limitations of current algorithms and improve accuracy. The algorithm shows superior performance in simulations and field tests.
Article
Robotics
Xiaoji Niu, Hailiang Tang, Tisheng Zhang, Jing Fan, Jingnan Liu
Summary: IC-GVINS is an INS-centric global navigation satellite system that achieves robustness and accuracy in complex environments.
IEEE ROBOTICS AND AUTOMATION LETTERS
(2023)
Article
Automation & Control Systems
Xinyu Ye, Jiayi Ma
Summary: This article proposes an effective and efficient visual place recognition (VPR) approach that integrates semantic, sequential, and spatial geometric information. The focus is on candidate selection and geometric verification, rather than feature extraction. The proposed method, neighborhood manifold preserving matching (NMP), utilizes sequence partitioning and sequence-to-sequence matching to improve VPR performance. Experimental results demonstrate the superiority of the proposed VPR method and its potential for integration with other pipelines.
IEEE TRANSACTIONS ON INDUSTRIAL INFORMATICS
(2023)
Article
Engineering, Electrical & Electronic
Jing Dong, Xingyu Ren, Songlai Han, Shilin Luo
Summary: This paper proposes an approach to aid autonomous navigation without external GNSS aiding using vision positioning carried by a UAV. The method significantly improves navigation performance through data fusion and image matching.
IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY
(2022)
Article
Engineering, Electrical & Electronic
Tianxiao Song, Yu Chen, Xianmu Li, Zhaoyang Li, Li Xing
Summary: This paper proposes an innovative fusion map matching method to improve the accuracy of low-cost ground autonomic positioning systems, and discusses the impact of inertial positioning errors on map matching.
IEEE SENSORS JOURNAL
(2022)
Article
Computer Science, Information Systems
Carlo Masone, Barbara Caputo
Summary: Visual place recognition (VPR) has become a topic of interest across multiple research communities, with a focus on the recent advances in deep learning. Research in this area covers topics such as image representations, metric learning techniques, and applications in robotics. The survey also provides an overview of various datasets used in VPR with different characteristics.
Article
Computer Science, Artificial Intelligence
Yatong Zhou, Ya Guo, Kuo-Ping Lin, Fan Yang, Lingling Li
Summary: This study addresses the issues of limited or unavailable UAV image matching labels, differences in imagery between UAV and ground cameras, and limited UAV airborne computer memory. To tackle these challenges, a novel unsupervised UAV image matching network called USuperGlue is developed, which utilizes local self-attention. The aim is to enhance the precision and recall of the matching process. The experiment demonstrates that the trained USuperGlue model achieves superior precision and recall, making it a highly effective solution for UAV image matching.
Article
Chemistry, Multidisciplinary
Xiaojie Liu, Xiaoting Guo, Donghua Zhao, Chong Shen, Chenguang Wang, Jie Li, Jun Tang, Jun Liu
APPLIED SCIENCES-BASEL
(2019)
Article
Chemistry, Analytical
Qing Lu, Lixin Pang, Haoqian Huang, Chong Shen, Huiliang Cao, Yunbo Shi, Jun Liu
Article
Chemistry, Analytical
Huiliang Cao, Yu Liu, Zhiwei Kou, Yingjie Zhang, Xingling Shao, Jinyang Gao, Kun Huang, Yunbo Shi, Jun Tang, Chong Shen, Jun Liu
Article
Chemistry, Multidisciplinary
Min Zhu, Lixin Pang, Zhijun Xiao, Chong Shen, Huiliang Cao, Yunbo Shi, Jun Liu
APPLIED SCIENCES-BASEL
(2019)
Article
Chemistry, Multidisciplinary
Yahui Peng, Xiaochen Liu, Chong Shen, Haoqian Huang, Donghua Zhao, Huiliang Cao, Xiaoting Guo
APPLIED SCIENCES-BASEL
(2019)
Article
Chemistry, Analytical
Tianqi Guo, Wenqiang Wei, Qi Cai, Rang Cui, Chong Shen, Huiliang Cao
Summary: This paper presents a new type of three-axis gyroscope that is driven by capacitive drive. The motion equation, capacitance design, and spring design of the gyroscope are introduced and corresponding formulas are derived. The gyroscope exhibits a larger bandwidth, higher mechanical sensitivity, and good impact resistance.
Article
Computer Science, Information Systems
Xiaolin Guo, Rang Cui, Shaochen Yan, Qi Cai, Wenqiang Wei, Chong Shen, Huiliang Cao
Summary: This paper studies a gyro structure of N = 3 Wineglass Mode Metal Cylindrical Resonator Gyroscope (WMMCRG). Compared with traditional Cylindrical Vibrating Gyroscope (CVG), the designed structure has higher scale factor and lower frequency split. A closed-loop controlling system with low error and low noise is designed for WMMCRG. Test results show that the bias instability, bias stability, zero bias, Angular Random Walk (ARW), and frequency split of WMMCRG are 1.974 degrees/h, 10.869 degrees/h, 10.3323 degrees/s, 16 (degrees)/root h, 0.02 Hz,respectively.
Article
Chemistry, Analytical
Chenguang Wang, Yuchen Cui, Yang Liu, Ke Li, Chong Shen
Summary: This paper proposes an accelerometer denoising method based on empirical mode decomposition (EMD) and time-frequency peak filtering (TFPF) to remove noise generated during the accelerometer calibration process. A new design of the accelerometer structure is introduced and analyzed, and an algorithm combining EMD and TFPF is proposed to deal with the noise. The algorithm effectively suppresses random noise and protects the characteristics of the original signal, with an error controlled within 0.5%.
Article
Chemistry, Analytical
Tianshang Zhao, Chenguang Wang, Chong Shen
Summary: This paper proposes a hybrid seamless MEMS-INS/MNS strategy to suppress drift and improve the navigation capability. By introducing deep self-learning and a Kalman filter, the proposed method successfully mitigates the heading accuracy error and improves robustness and computational efficiency.