Search

Category-level object pose estimation aims to predict the 6D object pose and size of arbitrary objects from known categories. It remains a challenge due to the large intra-class shape variation. Recently, the introduction of the shape prior adaptation mechanism into the normalized canonical coordinates (i.e., NOCS) reconstruction process has been shown to be effective in mitigating the intra-class shape variation. However, existing shape prior adaptation methods simply map the observed point cloud to the normalized object space, and the extracted object descriptors are not sufficient for the perception of the object pose. As a result, they fail to predict the pose of objects with complex geometric structures (e.g., cameras). To this end, this paper proposes a novel shape prior adaption method named MSSPA-GC for category-level object pose estimation. Specifically, our main network takes the observed instance point cloud converted from the RGB-D image and the prior shape point cloud pre-trained on the object CAD models as inputs. Then, a novel 3D graph convolution network and a PointNet-like MLP network are designed to extract pose-aware object features and shape-aware object features from these two inputs, respectively. After that, the two-stream object features are aggregated through a multi-scale feature propagation mechanism to generate comprehensive 3D object descriptors that maintain both pose-sensitive geometric stability and intra-class shape consistency. Finally, by leveraging object descriptors aware of both object pose and shape when reconstructing the NOCS coordinates, our approach elegantly achieves state-of-the-art performance on the widely used REAL275 and CAMERA25 datasets using only 25% of the parameters compared with existing shape prior adaptation models. Moreover, our method also exhibits decent generalization ability on the unconstrained REDWOOD75 dataset. [Display omitted] • A novel shape prior adaptation method for category-level object pose estimation. • A 3D-GCN to learn multi-scale pose-sensitive features from observed point clouds. • A MLP network to learn multi-scale shape-aware features from prior point clouds. • A feature propagation mechanism to fuse pose-aware features and shape-aware features. • State-of-the-art performance on the NOCS datasets with only 5.9M parameters. • Good generalization ability on the REDWOOD75 dataset without orientation constraints. [ABSTRACT FROM AUTHOR]

• We present CMA, a novel cross-modal data fusion approach that incorporates the attention mechanism. • CMA extracts discriminative cross-modal features which are more robust to 6D object pose estimation. • We evaluate our method on two widely used datasets: LINEMOD dataset and YCB-Video dataset. • Experimental results demonstrate that our method achieves superior performance on both datasets over the state-of-the-art methods as well as high-efficiency. [Display omitted] Deep learning methods for 6D object pose estimation based on RGB and depth (RGB-D) images have been successfully applied to robotic manipulation and grasping. Among these approaches, the fusion of RGB and depth modalities is one of the most critical issues. Most existing works performed fusion via either simple concatenation, or element-wise multiplication of the features generated by these two modalities. Despite achieving impressive progress, such fusion strategies do not explicitly consider the different contributions of RGB and depth modalities, leaving a gap for performance enhancement. In this paper, we present a Cross-Modal Attention (CMA) component for the problem of 6D object pose estimation. With the attention mechanism, features of two different modalities are aggregated adaptively through the attention weights, such that powerful representations from the RGB-D images can be efficiently extracted. Comprehensive experiments on both LINEMOD and YCB-Video datasets demonstrate that the proposed approach achieves state-of-the-art performance. [ABSTRACT FROM AUTHOR]

• We formulate the 2D joint locations of the human body as a graph and present a novel lightweight graph convolutional operation with structural knowledge about human bodies. • Based on the proposed graph convolutional operation, a novel graph-based mixture density network (GMDN) is proposed to resolve the ambiguity and occlusion problem existing in 3D human pose estimation. • Comprehensive experiments on the Human3.6M dataset demonstrate that GMDN achieves state-of-the-art performance with only 0.30M parameters. [Display omitted] 3D human pose estimation from 2D detections is an ill-posed problem because multiple solutions may exist due to the inherent ambiguity and occlusion. In this paper, we propose a novel graph-based mixture density network (GMDN) to tackle the 2D-to-3D human pose estimation problem. We formulate the 2D joint locations of the human body as a graph, and thus the pose estimation task can be redefined as a graph regression problem. Additionally, we present a novel graph convolutional operation with the incorporation of structural knowledge about human body configurations to assist with reasoning of the structural relations implied in the human bodies. Furthermore, we employ mixture density networks to formulate the 3D human poses as a multimodal distribution. The presented GMDN is lightweight with only 0.30M parameters, and the experimental results demonstrate that it achieves state-of-the-art performance. [ABSTRACT FROM AUTHOR]

Category-level 6D object pose estimation aims to predict the position and orientation of unseen object instances, which is a fundamental problem in robotic applications. Previous works mainly focused on exploiting visual cues from RGB images, while depth images received less attention. However, depth images contain rich geometric attributes about the object's shape, which are crucial for inferring the object's pose. This work achieves category-level 6D object pose estimation by performing sufficient geometric learning from depth images represented by point clouds. Specifically, we present a novel geometric consistency and geometric discrepancy learning framework called CD-Pose to resolve the intra-category variation, inter-category similarity, and objects with complex structures. Our network consists of a Pose-Consistent Module and a Pose-Discrepant Module. First, a simple MLP-based Pose-Consistent Module is utilized to extract geometrically consistent pose features of objects from the pre-computed object shape priors for each category. Then, the Pose-Discrepant Module, designed as a multi-scale region-guided transformer network, is dedicated to exploring each instance's geometrically discrepant features. Next, the NOCS model of the object is reconstructed according to the integration of consistent and discrepant geometric representations. Finally, 6D object poses are obtained by solving the similarity transformation between the reconstruction and the observed point cloud. Experiments on the benchmark datasets show that our CD-Pose produces superior results to state-of-the-art competitors. • A novel 3D geometry learning framework for category-level 6D object pose estimation • The network only relies on 3D point clouds to infer object poses without RGB features • A Pose-Consistent Module to extract the geometric consistency of each category • A Pose-Discrepant Module to exploit the geometric discrepancy of different instances • State-of-the-art performance on the NOCS-CAMERA and NOCS-REAL datasets [ABSTRACT FROM AUTHOR]

Abstract A three-dimensional high-order panel method based on Non-Uniform Rational B-Spline (NURBS) for predicting the ship–ship hydrodynamic interaction during meeting and overtaking in shallow water is developed. The NURBS surface is used to precisely represent the hull geometry. The velocity potential on the body surface is described by B-spline after the source density distribution on the boundary surface is determined. A collocation approach is applied to the boundary integral equation discretization, and the velocity potential is being solved at each time step. Under the low-speed assumption, the effect of free surface elevation is neglected in the numerical calculation, and the infinite image method is used to deal with the finite water depth effect. Two tankers in model tests are investigated, and the predicted hydrodynamic interaction forces and moments are compared with experimental measurements to clarify the validity of the proposed numerical method. Calculations are then conducted for different water depths, lateral ship–ship distances and ship speeds, and the detailed results are discussed to demonstrate the effects of these factors. Highlights • Ship–ship interaction in shallow water is investigated by a high-order panel method. • Interaction hydrodynamic forces during meeting and overtaking are predicted. • The method is validated by comparing the numerical results with experimental data. • Computations are conducted at different water depth, ship–ship distance and ship speed. • The influences of water depth, ship–ship distance and ship speed are demonstrated. [ABSTRACT FROM AUTHOR]

Abstract Renewable energy has been rapidly developing because of the increasing severity of pollution and energy crises. In particular, solar thermal power generation technology has been attracting considerable attention globally due to its large scale, low unit cost, and environment-friendliness. Thus, it has been quickly progressing. The intermittence and instability of solar energy can be solved by combination of solar thermal power with heat storage technology. Mixed molten salts stand out as heat transfer and storage materials due to their wide temperature range, low vapor pressure, high heat capacity, low viscosity, and economical environmental protection. In the present study, a new kind of quaternary nitrate molten salt was presented. The salt was Hitec salt with Ca(NO 3) 2 additive and has a low melting point, high decomposition point, and low cost. The thermophysical properties of the proposed salt, such as melting point, decomposition point, crystallization point, specific heat, density, thermal conductivity, and thermal stability, were measured. Results showed that the melting, crystallization, and decomposition points of the Hitec salt with Ca(NO 3) 2 additive were 83.1 °C, 163.1 °C, and 628.5 °C, respectively. The proper working temperature range of the new salt was 200–565 °C, which was wider than those of Hitec salt (200–450 °C) and solar salt (290–565 °C). The average specific heat and thermal conductivity of the Hitec salt with Ca(NO 3) 2 additive were approximately 1.52 J/(g K) and 0.655 W/(m K), respectively, which showed better heat storage and heat transfer performance than did Hitec salt (1.40 J/(g K) specific heat, 0.350 W/(m K) thermal conductivity) and solar salt (1.50 J/(g K) specific heat, 0.520 W/(m K) thermal conductivity). An experimental study was conducted on the stability of mixed molten salts at 565 °C for 1200 h and 120 times for quenching/heating at 200–500 °C, and the Hitec salt with Ca(NO 3) 2 additive exhibited good thermal stability. Highlights • A new quaternary nitrate molten salt, Hitec salt with Ca(NO 3) 2 additive is presented. • Working temperature range of new salt is wider than that of Hitec and solar salts. • Hitec salt with Ca(NO 3) 2 additive shows better heat storage and transfer performance. • Hitec salt with Ca(NO 3) 2 additive exhibits good thermal stability. [ABSTRACT FROM AUTHOR]

It is essential to establish reliable method to predict the ship maneuvering motions in complex navigation environments for navigation efficiency and safety. In this paper, a novel hybrid nonparametric modeling method (R-LSTM) is developed by coupling Bidirectional Long Short-Term Memory (Bi-LSTM) with residual network and attention mechanism. Firstly, the prediction capability of the proposed hybrid method is evaluated in the predictions of zigzag maneuvers in calm water. The model test data of a KVLCC2 tanker model are utilized for modeling and validation, and the optimal parameters are determined by Tree-structured Parzen Estimator algorithm. The results demonstrate that the proposed method has higher prediction accuracy and better generalization ability than the classical LSTM method. On this basis, the online modeling method for predicting the turning motion in regular waves is developed along with sliding time window and time series split cross-validation. The model test data of a KCS container ship and a ONRT tumblehome ship under different wave conditions are used for online modeling and real-time prediction. The results indicate that the ship trajectories with trajectory drift and fluctuation features during the turning motion under the wave impacts are captured well by the developed method with satisfactory confidence intervals. • A novel hybrid nonparametric modeling method (R-LSTM) is developed combining bidirectional long short-term memory model, residual network and attention mechanism. • The black-box modeling by R-LSTM is conducted to predict ship maneuvering motions in calm water and in regular waves. • R-LSTM has higher accuracy and better generalization than the classical LSTM in the predictions of zigzag maneuvers in calm water. • Sliding time window and time series split cross-validation are coupled with R-LSTM for online training and validation of nonparametric model of turning maneuvers in regular waves. • Ship turning trajectories with trajectory drift and fluctuation features under the wave impacts are captured well by the developed method. [ABSTRACT FROM AUTHOR]

Vortex-induced motions (VIM) is becoming a noteworthy issue for column-stabilized floating platforms, mainly due to its substantial fatigue damage to risers and mooring system. The VIM of deep-draft semi-submersible is more complex than single column floaters because of the wake interference between columns, as well as the considerable yaw motions. In the present work, a numerical approach for simulating VIM of deep-draft semi-submersible is proposed. Specifically, detached-eddy simulation is used for turbulence modeling and dynamic overset grid technique is used for moving objects. Simulations for stationary drag and VIM of a model-scale paired-column semi-submersible are conducted with the proposed approach. The numerical results are compared with experimental data. Transverse, in-line and yaw motions are allowed during VIM simulations and are further analyzed in frequency domain by Fast Fourier Transform (FFT). Different VIM characteristics are observed at different current velocities. The work done by each component of the structure is also discussed. Flow visualizations are presented for better understanding of the wake interferences during VIM. The accuracy and reliability of the current numerical approach is assessed. [ABSTRACT FROM AUTHOR]

The twin-screw fully appended ONR Tumblehome ship model is numerically studied for the standard 10/10 zigzag maneuver. Simulations are carried out using naoe-FOAM-SJTU, a CFD solver developed on open source platform, OpenFOAM. Overset grid technique and 6DoF module with a hierarchy of bodies are applied to solve the motions of the free-running ship with twin rotating propellers and turning rudders. The self-propulsion point is using the constant value obtained by the previous CFD simulation of the twin-screw self-propelled ship in calm water. Simulations of the standard zigzag maneuver in both calm water and different incident waves are considered. The present CFD results agree well with the available experimental measurement for both ship motions and the main parameters of zigzag maneuver. This indicates that the direct simulations for ship zigzag maneuver in waves utilizing present methods are feasible and reliable. In addition, it is found that unsteady effects of the hydrodynamic loads on the free-running ship during zigzag maneuver in waves is noticeable. Furthermore, flow visualizations in different incident waves are presented so as to give a better understanding of the wave influences on free-running ship maneuver. [ABSTRACT FROM AUTHOR]

This paper describes numerical simulations of the vortex-induced vibrations (VIVs) of a long flexible riser in a step current. We consider the model vertical riser tested at the Delta Flume. The simulation is carried out by our in-house computational fluid dynamics (CFD) solver viv-FOAM-SJTU, which was coupled with the strip method and developed in OpenFOAM platform. The vibration modes in both in-line (IL) and cross-flow (CF) directions are accurately predicted. The numerically predicted maximum mean IL displacement and its location differed marginally from the experimental results. The good agreement between the numerical and experimental results proved that this solver was reliable for predicting the VIV response. A large number of numerical tests were then carried out to study the effects of various parameters on VIV responses further. Three main parameters are considered in this study: current velocity, top tension and mass ratio. The intrinsic relationship between the natural frequency and oscillating frequency was analyzed to explain the occurrence of the dominant mode. Based on the numerical results, the regular characteristics of the VIV response with the reduced velocity were pointed out. The curvatures and the maximum mean offset values were proportional to the squares of the reduced velocities. [ABSTRACT FROM AUTHOR]

The electrochemical properties of Li–O 2 batteries are mainly restricted by the kinetics of the OER (oxygen evolution reaction) before realizing high-rate and long-term cycling performance. Among various excellent OER catalysts, RuO 2 has shown excellent catalytic activity and good electrical conductivity. Here, a bifunctional catalyst with RuO 2 well dispersed on NiCo 2 O 4 nanosheets has been studied, showing a low OER Tafel slope of as low as 58 mV/decade in alkaline media. While assembled in Li–O 2 batteries with RuO 2 @ NiCo 2 O 4 cathode, a large specific discharge capacity as high as 17,633 mAh g −1 can be obtained at a current density of 200 mA g −1 with a coulombic efficiency of 91%. A long-term stability about 128 cycles can be achieved without much deterioration. The excellent electrochemical performance is strongly correlated with the nanosheets like structure for easy O 2 diffusion, electrolyte permeation, sufficient space provision of Li 2 O 2 deposition and cooperation of NiCo 2 O 4 and RuO 2 , which exhibits excellent ORR (oxygen reduction reaction) and OER catalytic activity respectively. [ABSTRACT FROM AUTHOR]

This study aims at a comprehensive investigation of passing ship effects on a moored ship in confined waterways with high blockage, referring to the new benchmark cases from the latest MASHCON conference. The whole process of a KCS container ship passing a moored Neo-Panamax container ship is simulated by combining the URANS method with the dynamic overset mesh technique. The hydrodynamic forces and moments, attitudes of the moored ship, and ship-generated waves are predicted. Firstly, the case with 10% UKC is simulated for validation. The results indicate that the numerical layout with a larger initial longitudinal distance between the two ships presents more accurate results agreeing well with the test data. Then, based on the validated method, the effects of tank width, water depth, lateral passing distance and passing ship speed are analyzed and are quantified by curve fittings in terms of the amplitudes of hydrodynamic quantities. It is found that these parameters have remarkable impacts on the moored ship, among which the passing speed is the most significant factor. In addition, the ship-generated drawdown dominates in the whole passing process, which causes prominent variations of the hydrodynamic quantities and induces pronounced pressure reductions on the moored ship. • Entire passing process in confined waterways is simulated by URANS method combined with dynamic overset mesh technique. • A larger initial longitudinal distance between the two ships in simulations presents better results. • Numerical methods are validated by comparisons with benchmark test data for a case with only 10%UKC. • Effects of important parameters are quantified by curve fittings and the ship speed is the most significant factor. • Primary wave system and drawdown dominate in the passing process. [ABSTRACT FROM AUTHOR]

The free running ONR Tumblehome ship model is numerically studied under course keeping control. Simulations are carried out using naoe-FOAM-SJTU, a CFD solver developed on open source platform OpenFOAM. Overset grid technique and 6DOF module with a hierarchy of bodies are applied to handle the free running ship motions with twin actual rotating propellers and moving rudders. Self-propulsion computation in calm water is first performed to achieve the approach speed ( U =1.11 m/s, Fr =0.2) using a proportional-integral controller. The obtained self-propulsion model point is then used for the free running computation under course keeping control in waves, where three regular waves are considered: head wave, bow quartering wave and beam wave. To fulfil the course keeping demand, a new course keeping control module is developed using feedback controller based on the CFD solver naoe-FOAM-SJTU. The predicted results are compared with the benchmark data in Tokyo 2015 CFD Workshop in ship hydrodynamics. Good agreements are achieved for both self-propulsion and course keeping tests, which shows that free running ship simulations especially in waves using the CFD approach coupled with newly developed course keeping module is feasible and reliable. Furthermore, flow visualizations are also presented to explain the hydrodynamic performance of ship hull-propeller-rudder interaction in waves under course keeping conditions. [ABSTRACT FROM AUTHOR]

Background Coronary artery disease (CAD) and aortic stenosis share pathophysiological mechanisms and risk factors. We evaluated the clinical utility of coronary computed tomography angiography (CTA) to identify CAD and revascularization requirement in patients with severe aortic stenosis considered for transcatheter aortic valve replacement (TAVR). Methods Consecutive patients without known CAD underwent calcium scoring, CTA and invasive coronary angiography (ICA). A second-generation dual-source CT scanner was used. ICA-quantitative coronary angiography (QCA) served as reference standard. CAD was reported using a lenient threshold of ≥50% and a stricter threshold of ≥70% diameter reduction. Findings of ≥70% diameter reduction and of high-risk CAD were used to predict revascularization. Results The study included 140 patients [68 males; 82.3 (7.7) years]. CAD defined by the 50% threshold on ICA was found in 58/140 (41%) patients. CAD by the 70% threshold was found in 23/140 (16%) patients. High-risk CAD was found in 16/140 (11%) patients. CTA and ICA had similar odd-ratios of 3.22 (1.26–8.23) and 4.62 (1.64–13.05), respectively, in predicting revascularization. Forty-two/140 (30%) patients had <400 Agatston calcium score, 98/140 (70%) patients had ≥400 calcium score. The diagnostic performance of CTA in the low calcium score group was better than the high calcium score group (AUC 0.81 vs. 0.63). Conclusion CTA remained questionable to rule-out CAD as gatekeeper to ICA in TAVR candidates who had severe coronary calcifications. In patients with less severe coronary calcifications, accounting for 30% of participants in this study, CTA may play a clinical role. [ABSTRACT FROM AUTHOR]

In this work, novel alternative graphene-like hybrids covalent with porphyrin (porphyrin-graphene-like hybrids) are designed, named TPP-gra(n)-a and TPP-gra(n)-b (n = 1–5), respectively. Moreover, the differences in their electronic structures, one-photon absorption (OPA) and two-photon absorption (TPA) properties are revealed by theoretical chemistry calculations. Our study shows that the rearrangement of carbon-atom layers of the hybrids can effectively enhance electronic delocalization, bringing about outstanding TPA properties. In addition to the increase of the graphene-like size, substituting porphyrin into different active-carbon of graphene-like molecules can cause better polarization of TPP-gra(n)-b series than TPP-gra(n)-a series, resulting in TPP-gra(n)-b series possess relatively higher TPA cross sections ( δ TPA ) than that of corresponding TPP-gra(n)-a compounds, such as TPP-gra3-b, TPP-gra4-b and TPP-gra5-b, whose δ TPA can be as large as 5686.2–19496.6 GM ranging from 535.8 to 579.4 nm. The carbon-based materials possessing relatively high δ TPA values have a great potential for applications in optoelectronic devices and optical limiting fields. Consequently, we expect the calculations can provide a theoretical perspective for further studies on TPA materials based on graphene. [ABSTRACT FROM AUTHOR]

In the present study, we have designed a series of novel two-photon mercury ion (Hg 2+ ) fluorescent probes with large two-photon absorption cross-sections ( δ max ) based on intramolecular charge transfer (ICT), and carried out the quantum-chemical calculations on the electronic structural and two-photon absorption (TPA) properties of these probes. The calculated results reveal that the fluorescent peaks of the reaction products with Hg 2+ display significant red-shifts compared with these corresponding Hg 2+ probes, demonstrating that these two-photon Hg 2+ probe molecules can be applied as excellent ratiometric fluorescent bio-imaging reagents. More importantly, some designed molecules possess large δ max values (1000–3100 GM) in the near-infrared region (650–1100 nm), which can be viewed as more promising candidates for detecting Hg 2+ in the vivo biological system. Furthermore, we hope the investigations can provide a theoretical basis for developing novel two-photon fluorescent Hg 2+ probes in the biological system. [ABSTRACT FROM AUTHOR]

The high overpotentials of oxygen reduction reaction (ORR) and oxygen evolution reaction (OER) in the air cathode are still big challenge to be overcome for the high performance non-aqueous lithium-air batteries (LABs), especially in ambient air. In this paper, dandelion-like NiCo 2 O 4 hollow microspheres are prepared through a simple self-assembled hydrothermal process as cathode catalyst. The assembled LABs show relative high discharge potential in different oxygen partial pressure, and can deliver a reversible capacity of 8019 mAh·g −1 at the current density of 300 mA·g −1 with stable cycle performance for at least 40 times in ambient air. The excellent performance of NiCo 2 O 4 cathode can be attributed to the dandelion hollow structure, together with the abundant oxygen vacancies, which would enhance the ORR and OER process. The study confirms the feasibility of dandelion-like NiCo 2 O 4 hollow spheres as efficient cathode for the LABs in ambient air. [ABSTRACT FROM AUTHOR]

The electronic structures and photophysical properties of a series of the neutral and ionic Cu(I) complexes with 2-(2′-quinolyl)benzimidazole and phosphine mixed ligand were investigated using density functional theory (DFT) and time-dependent DFT (TD-DFT) methods. To investigate the differences between the neutral and ionic complexes, the highest occupied molecular orbital (HOMO) energy, the lowest virtual molecular orbital (LUMO) energy, Δ E H-L (the energy gap between HOMO and LUMO), the ionization potentials (IPs), electron affinities (EAs) and reorganization energies (λ) were computed. To explain the reason of the metal-to-ligand charge transfer (MLCT) hardly detection in the neutral complexes, the absorption and emission spectra were studied in detail. To evaluate the photoluminescence quantum yield (PLQY), the non-radiative and radiative decay rate constants ( k nr and k r ) were also presented. As a result, these calculations reveal that the structural distortion between the ground and excited states plays an important role in governing the PLQY. The charge transfer and the transport equilibrium property were markedly improved due to the lack of mobile counterions in the neutral form. Introducing the ether linkage together with the enhanced NˆN π-conjugation in 2b had obviously faster k r but slower k nr , which led to its higher PLQY relative to the introducing the ether linkage only or lacking of a proton and the counterions in complexes. [ABSTRACT FROM AUTHOR]

Summary Hantaviruses infect their reservoir hosts and humans, but the infection only causes disease in humans. In Asia and Europe (the Old World), the hantaviruses usually cause haemorrhagic fever with renal syndrome (HFRS). This article summarizes the current understanding of hantavirus epidemiology, as well as the clinical manifestations, pathogenesis, renal pathology, diagnosis, treatment, and prevention of HFRS. Moreover, the spatiotemporal distribution of HFRS was analysed based on the latest data obtained from the Chinese Centre for Disease Control and Prevention, for the period January 2004 to April 2015, to provide valuable information for the practical application of more effective HFRS control and prevention strategies in China. [ABSTRACT FROM AUTHOR]