315 research outputs found
Highly stretchable/soft silicone elastomers with tailored network structures
No full textThere is growing interest in developing stretchable/soft silicone elastomers for applications in advanced devices, including soft robotics, stretchable electronics, medical devices, and microfluidics. High stretchability permits various distortion scenarios and exceptional deformations, enabling wider applications of the devices. Besides, preparing silicone elastomers with a combined softness and elasticity resembling that of human soft tissue has been of great interest for use in soft robotics. Although various strategies have been developed for stretchable and soft silicone elastomers, none of them are versatile enough to enable the preparation of silicone elastomers that are both highly stretchable and very soft. This work aims to develop routes for molecular design of highly stretchable/soft silicone elastomers and to apply the resulting elastomers in one of the most promising fields, i. e. dielectric elastomer actuators (DEAs).Firstly, elastomers, cured by forming concatenated rings, were substantiated through a series of experiments. In contrast to conventional silicone elastomers which are crosslinked, the novel silicone elastomers are prepared from heterobifunctional PDMS macromonomers without the use of crosslinkers. Size exclusion chromatography of extracts confirmed the formation of non-concatenated monocyclic PDMS. Swelling experiments confirmed mechanically stable networks and high swelling ratios. Linear viscoelasticity measurements also demonstrated a behavior different from both PDMS melts and conventional crosslinked networks. The observed properties were explained by a formation of a network of concatenated rings which give rise to a continuous structure with a large degree of flexibility. Secondly, a facile curing route was developed to tailor the stretchability and the softness of elastomersby creating highly entangled elastomers and bottle-brush elastomers. The curing chemistry was based onrealizing that silicone elastomers form when telechelic/multiple Si-H functional PDMS and platinumcatalyst are heated to 100℃ in air. This observation was explained through crosslinking reactions of the Si-H functional groups with otherwise inert constituents of the PDMS chain in the presence of oxygen (SiHcrosslinking) as elucidated by subsequent mechanistic studies. Combining the hydrosilylation and SiH crosslinking reactions in a one-pot reaction allowed formation of highly entangled elastomers and bottlebrush elastomers from commercial precursor polymers. The highly entangled elastomers showed high stretchability with maximum strains of 2800%, and the bottle-brush elastomers exhibited extremely lowsoftness with shear moduli of 1.2-7.4 kPa. The reported curing chemistry was used to prepare a range of silicone elastomers with carefully tailored mechanical properties.To demonstrate the applications of the prepared materials, a highly stretchable silicone elastomer was applied in DEAs, aiming to decrease operation voltages by using thin prestretched films. The fabricated DEAs could be actuated to a 30% lateral strain at 4.3 kV for a 122 μm thick prestretched film, and to a 2.5% lateral strain at only 250 V for a 6.9 μm thick prestretched film. Lifetime and response speed tests further show that the DEAs are promising for applications where fast response speed is not strictly required. <br/
Modelling Fluid-structure Interaction in Offshore Photovoltaics
No full textThe main aim of the research presented in this report is investigating analytical methods to model fluid-structure interaction in large-scale offshore floating photovoltaics. The model that was attempted to be solved analytically is based on a model presented by Pengpeng Xu (2022).The dimensions in the equations were removed. Applying a perturbation method yielded hierarchic partial differential equations by introducing the wave amplitude divided by the depth of the ocean as a small perturbation parameter. The analytical solution of the first order problem was found by applying separation of variables and by using a Fourier transform. For certain classes of problems it is shown in this report that it is possible to analytically solve a model for fluid-structure interaction in offshore solar farms for various initial conditions.Applied Mathematic
Lithological Controls on Soil Aggregates and Minerals Regulate Microbial Carbon Use Efficiency and Necromass Stability
Get PDFMicrobial carbon (C) use efficiency (CUE) drives soil C formation, while physical-chemical protection stabilizes subsequent microbial necromass, both shaped by soil aggregates and minerals. Soils inherit many properties from the parent material, yet the influence of lithology and associated soil geochemistry on microbial CUE and necromass stabilization remains unknow. Here, we quantified microbial CUE in well-aggregated bulk soils and crushed aggregates, as well as microbial necromass in bulk soils and the mineral-associated organic matter fraction, originating from carbonate-containing (karst) and carbonate-free (clastic rock, nonkarst) parent materials along a broad climatic gradient. We found that aggregate crushing significantly increased microbial CUE in both karst and nonkarst soils. Additionally, compared to nonkarst soils, calcium-rich karst soils increased macroaggregate stability and decreased the ratio of oligotrophic to copiotrophic microbial taxa, leading to a reduction in microbial CUE. Moreover, microbial CUE was negatively associated with iron (hydr)oxides in karst soils, attributed to the greater abundance of iron (hydr)oxides and higher soil pH. Despite the negative effects of soil aggregation and minerals on microbial CUE, particularly in karst soils, these soils concurrently showed greater microbial necromass stability through organo-mineral associations compared to nonkarst soils. Consequently, (i) bedrock lithology mediates the effects of aggregates and minerals on microbial CUE and necromass stability; and (ii) balancing minerals’ dual roles in diminishing microbial CUE and enhancing microbial necromass stability is vital for optimizing soil C preservation
Numerical Simulation of the Interaction of A Membrane with Water with A Free Surface: Simulation of An Experiment by L. Rizos
No full textIn order to collect validation data for the study of the mechanism of fluid-structure interaction (FSI), an lab experiment was conducted by L. Rizos in the towing tank, 3ME,TUDelft in 2016. The concept of the experiment is shown in the figure 1. A cylindrical container is partially filled with water. A small cylindrical oscillator with flexible bottom is placed in the container. The oscillator is driven harmonically by a motor. During the experiment, the deflection of the flexible bottom, the motion of the free surface and the driven force were monitored and recorded. The figure 2 is the photo of the experiment. In order to better understand Rizos’ experiment, a series of researches are conducted in the Section Ship Hydromechanics, which includes analytical simulation and several numerical simulations with different methods. A linear algorithm is developed in this thesis, which applies implicit, monolithic (solving the fluid domain and structure domain simultaneously) and one-step (without iteration) methods. The model of the numerical simulation is shown in the figure 3, a small cylinder with flexible bottom is placed in the big cylindrical container. The two cylinders are partially filled with water and the still water levels are the same. The inner cylinder does not moves up and down as a whole. The oscillation of the whole system is the result of the initial wave elevation in the fluid domain and/or the initial deflection in the structure domain. The result of the numerical simulation is shown in the figure 4. The effect of added mass for a structure submerged in water results in that smaller eigen frequency of the structural vibration. The structure interacts with the ambient fluid, especially the free surface. For a pre-defined initial condition, the influence of the free surface results in the mode dispersion. The numerical periods of this FSI system agrees with the analytical periods. Significant numerical dissipation exists in the 1st order time integration techniques. Thus, the second order implicit Adam Moulton method, i.e., the trapezoidal rule, is implemented to improve the algorithm. in this way, the numerical dissipation is decreased drastically (5% less dissipation after 10 periods) without increasing the computational costs.Offshore and Dredging Engineerin
Alignbodynet: deep learning-based alignment of non-overlapping partial body point clouds from a single depth camera
Get PDFThis paper proposes a novel deep learning framework to generate omnidirectional 3D point clouds of human bodies by registering the front- and back-facing partial scans captured by a single depth camera. Our approach does not require calibration-assisting devices, canonical postures, nor does it make assumptions concerning an initial alignment or correspondences between the partial scans. This is achieved by factoring this challenging problem into ( i ) building virtual correspondences for partial scans, and ( ii ) implicitly predicting the rigid transformation between the two partial scans via the predicted virtual correspondences. In this study, we regress the SMPL vertices from the two partial scans for building the virtual correspondences. The main challenges are ( i ) estimating the body shape and pose under clothing from single partial dressed body point clouds, and ( ii ) the predicted bodies from front- and back-facing inputs required to be the same. We, thus, propose a novel deep neural network dubbed AlignBodyNet that introduces shape-interrelated features and a shape-constraint loss for resolving this problem.We also provide a simple yet efficient method for generating real-world partial scans from complete models, which fills the gap in the lack of quantitative comparisons based on the real-world data for various studies including partial registration, shape completion, and view synthesis. Experiments based on synthetic and real-world data show that our method achieves state-of-the-art performance in both objective and subjective terms
Deep learning-based 3D human body shape reconstruction from point clouds
No full text3D reconstruction of the human body shape is a fundamental problem in computer vision, which is valuable for various human-centric applications such as computer animation, virtual reality, and clothing design, to name a few. 3D scanning is a popular technology for acquiring the geometry of a subject based on which a 3D body reconstruction can be produced. Although countless body scanners were developed to meet different industrial requirements and a lot of advanced algorithms were proposed for optimizing the reconstructed body models, many problems are still not properly solved. These problems, however, are difficult to address using conventional methods.Recent years have witnessed the rapid development of artificial intelligence, especially deep learning. Encouraged by the significant success of deep learning in image processing, an increasing number of researchers attempted to extend deep learning to deal with 3D data. Following this trend, we proposed deep learning based solutions to several challenges existing in modern 3D body scanning and reconstruction.In this thesis, we focus on four challenges of 3D body scanning, namely, (i) estimation of body shape under clothing, (ii) body reconstruction from impaired point clouds, (iii) registration of non-overlapping point clouds, and (iv) animatable body reconstruction using a single depth camera. The first challenge arises from the fact that existing 3D scanning solutions require the subjects to get scanned with minimal clothing as the scanning device can only record the outmost surface of objects. This scanning procedure is inconvenient to most people and is also an infringement of the right to privacy. The second challenge comes from the observation that impaired point clouds are common in practice but they lack a systematic study. Moreover, the problems of misalignment and problematic posture are neglected in existing solutions. The third challenge is a classical problem: partial point cloud registration. We found that existing methods mainly rely on the assumption that the source and the target point clouds have sufficient overlap and none of them could handle non-overlapping registration. The last challenge is addressed as many applications demand dynamic human body models. Traditional methods require expensive professional devices to produce such models.We have addressed these four challenges by leveraging the deep learning paradigm. Our first contribution is to propose the first deep learning-based method in the literature for estimating the body shape under clothing from a single 3D dressed body scan. To facilitate the proposed model, a novel dataset consisting of large-scale dressed body scans and corresponding ground-truth body shapes is proposed. Our second contribution is a novel deep learning approach for jointly reconstructing an accurate body mesh and normalizing the posture of the human body model from a low-quality body point cloud in arbitrary postures. It proposes to directly reconstruct high-fidelity body shapes from impaired point clouds instead of attempting to point cloud repairments. Our third contribution is the first deep learning-based method in the literature to align non-overlapping partial point clouds Using this method, an omnidirectional body can be obtained from only two nonoverlapping body scans. The last contribution in this thesis is to propose a novel deep learning-based method to reconstruct an animatable body shape from only two depth images and at the same time allow for large pose variations between the camera shots.Extensive experiments based on different datasets have demonstrated that the proposed methods outperform the reference methods from the literature. Our work has resulted in numerous high-quality scientific publications and has demonstrated impact at both academic and industrial levels
Author response
No full textThe diverse cell types and the precise synaptic connectivity between them are the cardinal features of the nervous system. Little is known about how cell fate diversification is linked to synaptic target choices. Here we investigate how presynaptic neurons select one type of muscles, vm2, as a synaptic target and form synapses on its dendritic spine-like muscle arms. We found that the Notch-Delta pathway was required to distinguish target from non-target muscles. APX-1/Delta acts in surrounding cells including the non-target vm1 to activate LIN-12/Notch in the target vm2. LIN-12 functions cell-autonomously to up-regulate the expression of UNC-40/DCC and MADD-2 in vm2, which in turn function together to promote muscle arm formation and guidance. Ectopic expression of UNC-40/DCC in non-target vm1 muscle is sufficient to induce muscle arm extension from these cells. Therefore, the LIN-12/Notch signaling specifies target selection by selectively up-regulating guidance molecules and forming muscle arms in target cells. DOI:http://dx.doi.org/10.7554/eLife.00378.001
Motion based cable integrity limits for quadrant assisted pull-in operations on submarine inter-array cables
No full textThe installation of subsea cables connecting offshore wind turbines to the grid is a delicate process. This is especially the case for the operation of connecting the second end of the cable to the turbine. The applied method of using a quadrant means that in the workability analyses, multibody dynamics, line dynamics and sea state dynamics need to be combined, resulting in lengthy simulation requirements. The objective of this thesis is to determine vessel motion based limits to cable integrity in order to simplify workability analyses. This method allows the problem to be analyzed in the frequency domain, resulting in computational efficiency gains. In order to arrive at the desired result a literature study is performed regarding cable loading and cable failure modes encountered during quadrant assisted cable pull-ins. On that basis a detailed investigation into the relations between vessel motion and mechanical cable responses is carried out. To achieve this, a representative cable and a set of high but realistically encountered sea states are simulated. The obtained relations are then compared to the cable integrity limits for curvature, tension and compression to acquire limits expressed in terms of motion parameters such as acceleration, velocity and displacement. The results from these simulations show that: 1) maximum cable tension is closely correlated to upward heave velocity of the crane tip, 2) maximum cable compression is closely correlated to downward heave velocity, 3) maximum curvature is most closely correlated to downward heave velocity. It is concluded from the results that the cable response can be determined from the crane tip heave motion, which in turn is known from the vessel motions. This means that analysis of such a problem is possible in the frequency domain. As the results show that heave motion is governing in cable failure, heave compensation in the crane is recommended for the operations considered. In addition, an enhancement of the analysis process is proposed by extracting the linear relations and vessel motion limits from a small set of time domain simulations and assessing the situation thoroughly in a frequency domain analysis. The configurations considered exclude any effects added by cable protection systems or interaction effects with rigid bodies in the vicinity of the cable. Even though the analysis process is generally applicable and constitutes a significant improvement in computational load, the obtained relations may be generalized further by implementing a closed formula relating vessel motion to cable failure, or application of the cable protection system to the assessed configuration. Additional research in these directions is encouraged.Offshore and Dredging Engineerin
Deep learning-based 3D human body shape reconstruction from point clouds
No full text3D reconstruction of the human body shape is a fundamental problem in computer vision, which is valuable for various human-centric applications such as computer animation, virtual reality, and clothing design, to name a few. 3D scanning is a popular technology for acquiring the geometry of a subject based on which a 3D body reconstruction can be produced. Although countless body scanners were developed to meet different industrial requirements and a lot of advanced algorithms were proposed for optimizing the reconstructed body models, many problems are still not properly solved. These problems, however, are difficult to address using conventional methods.Recent years have witnessed the rapid development of artificial intelligence, especially deep learning. Encouraged by the significant success of deep learning in image processing, an increasing number of researchers attempted to extend deep learning to deal with 3D data. Following this trend, we proposed deep learning based solutions to several challenges existing in modern 3D body scanning and reconstruction.In this thesis, we focus on four challenges of 3D body scanning, namely, (i) estimation of body shape under clothing, (ii) body reconstruction from impaired point clouds, (iii) registration of non-overlapping point clouds, and (iv) animatable body reconstruction using a single depth camera. The first challenge arises from the fact that existing 3D scanning solutions require the subjects to get scanned with minimal clothing as the scanning device can only record the outmost surface of objects. This scanning procedure is inconvenient to most people and is also an infringement of the right to privacy. The second challenge comes from the observation that impaired point clouds are common in practice but they lack a systematic study. Moreover, the problems of misalignment and problematic posture are neglected in existing solutions. The third challenge is a classical problem: partial point cloud registration. We found that existing methods mainly rely on the assumption that the source and the target point clouds have sufficient overlap and none of them could handle non-overlapping registration. The last challenge is addressed as many applications demand dynamic human body models. Traditional methods require expensive professional devices to produce such models.We have addressed these four challenges by leveraging the deep learning paradigm. Our first contribution is to propose the first deep learning-based method in the literature for estimating the body shape under clothing from a single 3D dressed body scan. To facilitate the proposed model, a novel dataset consisting of large-scale dressed body scans and corresponding ground-truth body shapes is proposed. Our second contribution is a novel deep learning approach for jointly reconstructing an accurate body mesh and normalizing the posture of the human body model from a low-quality body point cloud in arbitrary postures. It proposes to directly reconstruct high-fidelity body shapes from impaired point clouds instead of attempting to point cloud repairments. Our third contribution is the first deep learning-based method in the literature to align non-overlapping partial point clouds Using this method, an omnidirectional body can be obtained from only two nonoverlapping body scans. The last contribution in this thesis is to propose a novel deep learning-based method to reconstruct an animatable body shape from only two depth images and at the same time allow for large pose variations between the camera shots.Extensive experiments based on different datasets have demonstrated that the proposed methods outperform the reference methods from the literature. Our work has resulted in numerous high-quality scientific publications and has demonstrated impact at both academic and industrial levels
Deep learning-based 3D human body shape reconstruction from point clouds
No full text3D reconstruction of the human body shape is a fundamental problem in computer vision, which is valuable for various human-centric applications such as computer animation, virtual reality, and clothing design, to name a few. 3D scanning is a popular technology for acquiring the geometry of a subject based on which a 3D body reconstruction can be produced. Although countless body scanners were developed to meet different industrial requirements and a lot of advanced algorithms were proposed for optimizing the reconstructed body models, many problems are still not properly solved. These problems, however, are difficult to address using conventional methods.Recent years have witnessed the rapid development of artificial intelligence, especially deep learning. Encouraged by the significant success of deep learning in image processing, an increasing number of researchers attempted to extend deep learning to deal with 3D data. Following this trend, we proposed deep learning based solutions to several challenges existing in modern 3D body scanning and reconstruction.In this thesis, we focus on four challenges of 3D body scanning, namely, (i) estimation of body shape under clothing, (ii) body reconstruction from impaired point clouds, (iii) registration of non-overlapping point clouds, and (iv) animatable body reconstruction using a single depth camera. The first challenge arises from the fact that existing 3D scanning solutions require the subjects to get scanned with minimal clothing as the scanning device can only record the outmost surface of objects. This scanning procedure is inconvenient to most people and is also an infringement of the right to privacy. The second challenge comes from the observation that impaired point clouds are common in practice but they lack a systematic study. Moreover, the problems of misalignment and problematic posture are neglected in existing solutions. The third challenge is a classical problem: partial point cloud registration. We found that existing methods mainly rely on the assumption that the source and the target point clouds have sufficient overlap and none of them could handle non-overlapping registration. The last challenge is addressed as many applications demand dynamic human body models. Traditional methods require expensive professional devices to produce such models.We have addressed these four challenges by leveraging the deep learning paradigm. Our first contribution is to propose the first deep learning-based method in the literature for estimating the body shape under clothing from a single 3D dressed body scan. To facilitate the proposed model, a novel dataset consisting of large-scale dressed body scans and corresponding ground-truth body shapes is proposed. Our second contribution is a novel deep learning approach for jointly reconstructing an accurate body mesh and normalizing the posture of the human body model from a low-quality body point cloud in arbitrary postures. It proposes to directly reconstruct high-fidelity body shapes from impaired point clouds instead of attempting to point cloud repairments. Our third contribution is the first deep learning-based method in the literature to align non-overlapping partial point clouds Using this method, an omnidirectional body can be obtained from only two nonoverlapping body scans. The last contribution in this thesis is to propose a novel deep learning-based method to reconstruct an animatable body shape from only two depth images and at the same time allow for large pose variations between the camera shots.Extensive experiments based on different datasets have demonstrated that the proposed methods outperform the reference methods from the literature. Our work has resulted in numerous high-quality scientific publications and has demonstrated impact at both academic and industrial levels
- …
