Neural-network-based models of 3-D objects for virtualized reality: a comparative study

The paper presents a comprehensive analysis and comparison of the representational capabilities of three neural architectures for three-dimensional (3-D) object representation in terms of purpose, computational cost, complexity, conformance and convenience, ease of manipulation, and potential applications in the context of virtualized reality. Starting from a pointcloud that embeds the shape of the object to be modeled, a volumetric representation is obtained using a multilayer feedforward neural network (MLFFNN) or a surface representation using either the self-organizing map (SOM) or the neural gas network. The representation provided by the neural networks (NNs) is simple, compact, and accurate. The models can be easily transformed in size, position, and shape. Some potential applications of the presented architectures in the context of virtualized reality are for the modeling of set operations and object morphing, for the detection of object collision, and for object recognition, object motion estimation, and segmentation.     

Micromechanical voltage reference using the pull-in of a beam

The pull-in voltage of a single-side anchored freestanding beam, under lateral deflection, has been investigated for application as a DC voltage reference. Two sets of electrodes, along side the tip, are used for parallel-plate type of electrostatic actuation of the 200 μm long beam in the plane of the wafer. Another set of buried electrodes is aligned with the plate electrode at the free-standing tip and is used as a differential capacitor for the simultaneous detection of the displacement, with the purpose to determine the stability border and thus the pull-in voltage. The single-end clamping ensures that the pull-in voltage is insensitive to technology-induced stresses. A 2D energy-based analytical model for the static pull-in is compared with measurements. Bifurcation diagrams are computed numerically, based on a local continuation method. Devices have been designed and fabricated in an epi-poly process. Measurements are in agreement with modeling and confirm a pull-in voltage in the 9.1-9.5 V range. Reproducibility is limited by hysteresis and charging of the dielectric layer in between the electrodes. The device can be operated in feedback or as a seesaw, by using the two sets of electrodes     

Bacterial Cellulose Hybrid Composites with Calcium Phosphate for Bone Tissue Regeneration

Bacterial cellulose (BC) is a unique microbial biopolymer with a huge number of significant applications in the biomedical field, including bone tissue engineering. The present study proposes to obtain and characterize BC hybrid composites with calcium phosphate as biocompatible and bioactive membranes for bone tissue engineering. BC precursor membranes were obtained in static culture fermentation, and after purification, were oxidized to obtain 2,3-dialdehyde bacterial cellulose (DABC). Calcium phosphate-BC oxidized membranes were produced by successive immersion in precursor solutions under ultrasonic irradiation. The samples were characterized for their physicochemical properties using scanning electron microscopy (SEM) coupled with energy-dispersive X-ray spectroscopy, attenuated total reflectance-Fourier transform infrared (ATR-FTIR) spectroscopy grazing incidence X-ray diffraction (GI-XRD), solid-state ¹³C nuclear magnetic resonance (CP/MAS ¹³C NMR), and complex thermal analysis. In vitro cell studies were also performed to evaluate the influence of modified morphological characteristics on cell adhesion and proliferation. The results showed an increase in porosity and biodegradability for DABC hybrid composites compared with BC. In vitro cell studies have revealed that both hybrid composites favor cell adhesion to the surface. The new BC and DABC hybrid composites with calcium phosphate could be considered promising materials for bone tissue regeneration.     

Approximate and Near Weak Invariance for Nonautonomous Differential Inclusions

Let X be a real Banach space and I a nonempty interval. Let \(K:I\rightsquigarrow X\) be a multi-function with the graph \(\mathcal {K} \). We give here a characterization for \(\mathcal {K} \) to be approximate/near weakly invariant with respect to the differential inclusion \(x^{\prime }(t)\in F(t, x(t))\) by means of an appropriate tangency concept and Lipschitz conditions on F. The tangency concept introduced in this paper extends in a natural way the quasi-tangency concept introduced by Cârj? et al. (Trans Amer Math Soc. [2009];361:343–90) (see also Cârj? et al. ([2007])). Viability, invariance and applications. Amsterdam: Elsevier Science B V) in the case when F is independent of t. As an application, we give some results concerning the set of solutions for the differential inclusion \(x^{\prime }(t)\in F(t,x(t))\).     

Robotic tactile recognition of pseudorandom encoded objects

This paper discusses an original model-based method for blind robotic tactile recognition of three-dimensional objects. Conveniently shaped geometric symbols representing terms of a pseudorandom array (PRA) are embossed on object surfaces. Symbols recovered by tactile probing are recognized using a neural network and then clustered in a PRA window that contains enough information to fully identify the absolute coordinates of the recovered window within the encoding PRA. By knowing how different object models were mapped to the PRA, it is possible to unambiguously identify the object face and the exact position of the recovered symbols on the face.     

New self‐normalized blind source separation networks for instantaneous and convolutive mixtures

The main blind source separation networks proposed in this paper apply to convolutive mixtures (including instantaneous ones). They have a recurrent or direct structure and they may use channel‐specific separating functions. They are based on a self‐normalized weight adaptation rule, which adaptively estimates the average powers of non‐linear functions of the network outputs. This allows us to control several aspects of the operation of these networks, especially their convergence speed/accuracy trade‐off. It also makes them more robust with respect to non‐stationary situations. We analyse their convergence properties. We validate all these results by means of experimental tests performed with these networks, classical ones, and additionally proposed linear instantaneous direct networks based on a normalization of their outputs. These tests especially show that the proposed networks improve the convergence trade‐off and that only these networks apply to highly mixed non‐stationary sources. Copyright © 2002 John Wiley & Sons, Ltd.     

Novel nanostructured pHEMA-TiO2 hybrid materials with efficient light-induced charge separation

We report on a new approach to the fabrication of an electronic material: organic-inorganic pHEMA-oxo-TiO(2) hybrid with efficient light-induced separation of charges. Particular attention is paid to the material nanoscale morphology. The size-selected 5.0 nm titanium oxo-alkoxy nanoparticles are prepared in a sol-gel reactor with rapid (turbulent) fluid micromixing and the ligand exchange results in a stable nanoparticulate precursor in HEMA solution, in which polymerization can be induced thermally or by photons. The obtained hybrid materials demonstrate the highest quantum yield of photoinduced charge separation of 50% and can store photoinduced electrons at a number density above 10% Ti atoms.     

3D-TEM investigation of the nanostructure of a δ-Al2O3 catalyst support decorated with Pd nanoparticles

The electron tomography technique applied in a quantitative way allowed us to characterize a heterogeneous catalyst made of Pd nanoparticles deposited on a δ-Al(2)O(3) lamellar support. In the first step, high resolution tomographic experiments carried out on several typical areas of support have confirmed the hypothesis of formation of δ-Al(2)O(3) proposed in the literature by the coalescence of lateral facets of the γ-Al(2)O(3) precursor. A bimodal porosity was also observed in the arrangement of δ-Al(2)O(3) platelets. In the second step, the Pd nanoparticles were found preferentially anchored on the lateral facets of δ-Al(2)O(3) platelets or on the defects situated on their basal planes. From a general point of view, we have demonstrated once again that the electron tomography technique implemented with nanometre resolution provides unique insight into the structure, morphology and spatial arrangement of components in a complex 3D nanostructure.