Article
Computer Science, Artificial Intelligence
Kang Liu, Lingling Chen, Liang Xie, Jian Yin, Shuwei Gan, Ye Yan, Erwei Yin
Summary: This paper proposes an automatic calibration method for multi-camera systems without the need for a known calibration pattern. By using the human body as the counterpart of the calibration pattern, the authors demonstrate superior performance compared to traditional methods.
IET COMPUTER VISION
(2022)
Article
Chemistry, Analytical
Bo Gu, Jianxun Liu, Huiyuan Xiong, Tongtong Li, Yuelong Pan
Summary: The proposed registration algorithm ECPC-ICP integrates road information constraints for accurate vehicle pose estimation based on sparse point clouds. Experimental results show improved accuracy and robustness compared to current methods, demonstrating its effectiveness in practical applications.
Article
Engineering, Aerospace
Xiang Liu, Hongyuan Wang, Xinlong Chen, Weichun Chen, Zhengyou Xie
Summary: This article proposes a position awareness network (PANet) for spacecraft pose estimation, which solves the problem of low estimation accuracy by extracting key points and constructing local structural descriptors. The matching matrix between point clouds is calculated to solve the pose using weighted singular value decomposition (SVD). Experimental results demonstrate that PANet outperforms state-of-the-art methods.
IEEE TRANSACTIONS ON AEROSPACE AND ELECTRONIC SYSTEMS
(2023)
Article
Automation & Control Systems
Jin Wu, Ming Liu, Yulong Huang, Chi Jin, Yuanxin Wu, Changbin Yu
Summary: This article proposes a new technique called SE(n)++, which maps SE(n) to SO(n+1) and transforms the coupling between rotation and translation into a more unified formulation, leading to better analytical results and computational performances. Experimental validations have confirmed the effectiveness of this method in point-cloud registration, hand-eye calibration, and SE(n) synchronization.
IEEE TRANSACTIONS ON CYBERNETICS
(2022)
Article
Engineering, Electrical & Electronic
Yunhao Cui, Yi An, Wei Sun, Huosheng Hu, Xueguan Song
Summary: Bucket pose estimation is a key technology for intelligent mining excavators. Existing methods suffer from cumulative errors and local optima issues. To address the problem of forgetting due to sample imbalance, a memory-augmented registration network (MARNet) is proposed. Experimental results demonstrate that MARNet improves the estimation accuracy of bucket pose, contributing to the development of intelligent mining excavators.
IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT
(2022)
Article
Chemistry, Analytical
Yiping Shao, Zhengshuai Fan, Baochang Zhu, Jiansha Lu, Yiding Lang
Summary: A point cloud data-driven pallet pose estimation method using an active binocular vision sensor is proposed, which overcomes the shortcomings of traditional pose estimation methods and achieves efficient and accurate estimations for driverless industrial trucks. Experimental results show its superiority in terms of accuracy and feature extraction time.
Article
Environmental Sciences
Jie Li, Yiqi Zhuang, Qi Peng, Liang Zhao
Summary: A non-cooperative target pose measurement system fused with multi-source sensors was designed in this study, using a cross-source point cloud fusion algorithm and a plane clustering-based CGA to improve measurement accuracy. Numerical simulations and semi-physical experiments confirmed the performance of the system, showing that the proposed algorithm can achieve high registration accuracy in point clouds with different densities and small overlap rates.
Article
Engineering, Aerospace
Guohua Kang, Qi Zhang, Jiaqi Wu, Han Zhang
Summary: This paper presents a method for relative position and attitude estimation using consecutive point clouds without feature extraction, which resolves inaccurate state estimation problems for non-cooperative targets caused by mismatched point pairs or low tracking accuracy. Experimental results show that the method can achieve effective and continuous estimation of target motion state.
Article
Chemistry, Analytical
Philipp Middendorf, Richard Bluemel, Lennart Hinz, Annika Raatz, Markus Kaestner, Eduard Reithmeier
Summary: This article introduces a high-precision inspection method for turbine blades in confined spaces, utilizing borescopic inspection and feature matching technology for assessment and wear analysis, providing guidance for engine disassembly decisions.
Article
Computer Science, Artificial Intelligence
Sheng Yu, Di-Hua Zhai, Yuyin Guan, Yuanqing Xia
Summary: This paper proposes a category-level object pose estimation network trained on synthetic data but capable of delivering good performance on real datasets. By introducing fusion and attention modules, the network enhances prediction accuracy. Through self-supervised learning and a small amount of real data supplementation, the method achieves high-precision pose estimation in real-world scenes.
IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS
(2023)
Article
Geography, Physical
Zhenghua Zhang, Guoliang Chen, Xuan Wang, Mingcong Shu
Summary: This paper proposes a learning-based network, Fore-Net, to filter outlier points and improve the quality of point clouds for subsequent tasks. By utilizing a dual space attention module and automatic labeling strategies, Fore-Net achieves high accuracy and robustness in various indoor scenes. It can enhance the performance of registration algorithms and denoise point clouds effectively.
ISPRS JOURNAL OF PHOTOGRAMMETRY AND REMOTE SENSING
(2022)
Article
Robotics
Dinh-Cuong Hoang, Johannes A. Stork, Todor Stoyanov
Summary: Estimating the 6DOF pose of objects is crucial for various applications, but it is challenging to do so accurately and quickly from 3D point clouds. This study proposes an end-to-end learning approach that utilizes a self-attention mechanism and deep Hough voting scheme to tackle this challenge.
IEEE ROBOTICS AND AUTOMATION LETTERS
(2022)
Article
Engineering, Electrical & Electronic
Guangliang Zhou, Deming Wang, Yi Yan, Huiyi Chen, Qijun Chen
Summary: The research proposes a semi-supervised pose estimation method using labeled synthetic data and unlabeled real data. It improves network performance through a self-supervised pipeline and feature mapping, and enhances accuracy with an attention-based pose estimation network.
IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS FOR VIDEO TECHNOLOGY
(2022)
Article
Chemistry, Analytical
Yu Gan, Guangmin Li, Guodong Liu, Binghui Lu
Summary: This paper proposes a relative-pose-measurement algorithm based on double-constrained intersurface mutual projections, which improves the accuracy of the matchings and achieves high-accuracy relative pose measurement by constructing the initial corresponding set and applying the rigid-transformation-consistency constraint.
Article
Engineering, Multidisciplinary
Xiaodong Lu, Ying Li
Summary: This paper proposes a new method for obtaining ship pose information estimation from multiple points based on point cloud data. By studying the spatial structure information of the ship and tracking the trajectory changes of key points, real-time angle, distance, velocity, and other information of the ship's multiple points are estimated. The results of the method verification show that it has smaller errors compared to other existing methods and provides more accurate and abundant ship pose information.
