Continual learning on nano drones

Miniaturized flying robotic platforms, called nano-drones, have the potential to revolutionize the autonomous robots industry sector thanks to their very small form factor. The nano-drones’ limited payload only allows for a sub-100mW microcontroller unit for the on-board computations. Therefore, traditional computer vision and control algorithms are too computationally expensive to be executed on board these palm-sized robots, and we are forced to rely on artificial intelligence to trade off accuracy in favor of lightweight pipelines for autonomous tasks. However, relying on deep learning exposes us to the problem of generalization since the deployment scenario of a convolutional neural network (CNN) is often composed by different visual cues and different features from those learned during training, leading to poor inference performances. Our objective is to develop and deploy and adaptation algorithm, based on the concept of latent replays, that would allow us to fine-tune a CNN to work in new and diverse deployment scenarios. To do so we start from an existing model for visual human pose estimation, called PULPFrontnet, which is used to identify the pose of a human subject in space through its 4 output variables, and we present the design of our novel adaptation algorithm, which features automatic data gathering and labeling and on-device deployment. We therefore showcase the ability of our algorithm to adapt PULP-Frontnet to new deployment scenarios, improving the R2 scores of the four network outputs, with respect to an unknown environment, from approximately [−0.2, 0.4, 0.0,−0.7] to [0.25, 0.45, 0.2, 0.1]. Finally we demonstrate how it is possible to fine-tune our neural network in real time (i.e., under 76 seconds), using the target parallel ultra-low power GAP 8 System-on-Chip on board the nano-drone, and we show how all adaptation operations can take place using less than 2mWh of energy, a small fraction of the available battery power.

A plasticidade dos amorfos: fazendo pigmentos brancos com fosfato de alumínio

A new white pigment made out of nano-structured non-crystalline aluminum phosphate was recently launched as an industrial product. Pigment opacifying properties are not intrinsic to aluminum phosphate but they arise as the result of a rare hollow particle nano-structure. This is in turn derived from the core-and-shell structure of amorphous aluminum phosphate precipitated under well-defined conditions. The new pigment is a product of the often neglected chemistry of non-crystalline ionic solids that can probably be a rich source of new successful products. The text describes a short account of the R&D activities, from the initial ideas to the present.

Influence of implant surfaces on osseointegration

The biological fixation between the dental implant surfaces and jaw bones should be considered a prerequisite for the long-term success of implant-supported prostheses. In this context, the implant surface modifications gained an important and decisive place in implant research over the last years. As the most investigated topic in, it aided the development of enhanced dental treatment modalities and the expansion of dental implant use. Nowadays, a large number of implant types with a great variety of surface properties and other features are commercially available and have to be treated with caution. Although surface modifications have been shown to enhance osseointegration at early implantation times, for example, the clinician should look for research evidence before selecting a dental implant for a specific use. This paper reviews the literature on dental implant surfaces by assessing in vitro and in vivo studies to show the current perspective of implant development. The review comprises quantitative and qualitative results on the analysis of bone-implant interface using micro and nano implant surface topographies. Furthermore, the perspective of incorporating biomimetic molecules (e.g.: peptides and bone morphogenetic proteins) to the implant surface and their effects on bone formation and remodeling around implants are discussed.

Study of the surface hardness and modulus of elasticity of conventional and microwave-cured acrylic resins

The aim of this study was to evaluate the following acrylic resins: Clássico®, QC-20® and Lucitone®, recommended specifically for thermal polymerization, and Acron MC® and VIPI-WAVE®, made for polymerization by microwave energy. The resins were evaluated regarding their surface nanohardness and modulus of elasticity, while varying the polymerization time recommended by the manufacturer. They were also compared as to the presence of water absorbed by the samples. The technique used was nanoindentation, using the Nano Indenter XP®, MTS. According to an intra-group analysis, when using the polymerization time recommended by the manufacturer, a variation of 0.14 to 0.23 GPa for nanohardness and 2.61 to 3.73 GPa for modulus of elasticity was observed for the thermally polymerized resins. The variation for the resins made for polymerization by microwave energy was 0.15 to 0.22 GPa for nanohardness and 2.94 to 3.73 GPa for modulus of elasticity. The conclusion was that the Classico® resin presented higher nanohardness and higher modulus of elasticity values when compared to those of the same group, while Acron MC® presented the highest values for the same characteristics when compared to those of the same group. The water absorption evaluation showed that all the thermal polymerization resins, except for Lucitone®, presented significant nanohardness differences when submitted to dehydration or rehydration, while only Acron MC® presented no significant differences when submitted to a double polymerization time. Regarding the modulus of elasticity, it was observed that all the tested materials and products, except for Lucitone®, showed a significant increase in modulus of elasticity when submitted to a lack of hydration.

Mechanical strength and thermal conductivity of low-porosity gypsum plates

The thermal conductivity and mechanical strength of gypsum and gypsum-cellulose plates made from commercial plaster by a new process have been measured. The gypsum parts made by the new process, 'novogesso', have high mechanical strength and low porosity. The gypsum strength derives from both the high aspect ratio of the gypsum crystals and the strong adhesion among them by nano-flat layers of confined water, which behaves as supercooled water. Another contribution to the strength comes from the nano-flatness of the lateral surfaces of the gypsum single crystals. The bending and compression strengths, σB and σc, of gypsum plates prepared by this new technique can be as high as 30 and 100 MPa, respectively. The way gypsum plates have been assembled as well as their low thermal conductivity allowed for the construction of a low-cost experimental house with thermal and acoustic comfort.

Preparation of PVP hydrogel nanoparticles using lecithin vesicles

Hydrogels micro, sub-micro and nanoparticles are of great interest for drug encapsulation and delivery or as embolotherapic agents. In this work it is described the preparation of nano and sub-microparticles of pre-formed, high molecular weight and monomer free poly(N-vinyl-2-pyrrolidone) encapsulated inside the core of lecithin vesicles. The hydrogel particles are formed with a very narrow diameter distribution, of about 800 nm, and a moderate swelling ratio, of approximately 10.

Preliminary biocompatibility investigation of magnetic albumin nanosphere designed as a potential versatile drug delivery system

Background: The magnetic albumin nanosphere (MAN), encapsulating maghemite nanoparticles, was designed as a magnetic drug delivery system (MDDS) able to perform a variety of biomedical applications. It is noteworthy that MAN was efficient in treating Ehrlich's tumors by the magnetohyperthermia procedure. Methods and materials: In this study, several nanotoxicity tests were systematically carried out in mice from 30 minutes until 30 days after MAN injection to investigate their biocompatibility status. Cytometry analysis, viability tests, micronucleus assay, and histological analysis were performed. Results: Cytometry analysis and viability tests revealed MAN promotes only slight and temporary alterations in the frequency of both leukocyte populations and viable peritoneal cells, respectively. Micronucleus assay showed absolutely no genotoxicity or cytotoxicity effects and histological analysis showed no alterations or even nanoparticle clusters in several investigated organs but, interestingly, revealed the presence of MAN clusters in the central nervous system (CNS). Conclusion: The results showed that MAN has desirable in vivo biocompatibility, presenting potential for use as a MDDS, especially in CNS disease therapy.

The Ultrastructural Study of Tumorigenic Cells Using Nanobiomarkers

Despite recent advances, patients with malignant brain tumors still have a poor prognosis. Glioblastoma (WHO grade 4 astrocytoma), the most malignant brain tumor, represents 50% of all astrocytomas, with a median survival rate of <1 year. It is, therefore, extremely important to search for new diagnostic and therapeutic approaches for patients with glioblastoma. This study describes the application of superparamagnetic nano-particles of iron oxide, as well as monoclonal antibodies, of immunophenotypic significance, conjoined to quantum dots for the ultrastructural assessment of glioblastoma cells. For this proposal, an immunophenotypic study by flow cytometry was carried out, followed by transmission electron microscopy analysis. The process of tumor cell labeling using nanoparticles can successfully contribute to the identification of tumorigenic cells and consequently for better understanding of glioblastoma genesis and recurrence. In addition, this method may help further studies in tumor imaging, diagnosis, and prognostic markers detection.

Turbulent magnetic pumping in a Babcock-Leighton solar dynamo model

Context. The turbulent pumping effect corresponds to the transport of magnetic flux due to the presence of density and turbulence gradients in convectively unstable layers. In the induction equation it appears as an advective term and for this reason it is expected to be important in the solar and stellar dynamo processes. Aims. We explore the effects of turbulent pumping in a flux-dominated Babcock-Leighton solar dynamo model with a solar-like rotation law. Methods. As a first step, only vertical pumping has been considered through the inclusion of a radial diamagnetic term in the induction equation. In the second step, a latitudinal pumping term was included and then, a near-surface shear was included. Results. The results reveal the importance of the pumping mechanism in solving current limitations in mean field dynamo modeling, such as the storage of the magnetic flux and the latitudinal distribution of the sunspots. If a meridional flow is assumed to be present only in the upper part of the convective zone, it is the full turbulent pumping that regulates both the period of the solar cycle and the latitudinal distribution of the sunspot activity. In models that consider shear near the surface, a second shell of toroidal field is generated above r = 0.95 R(circle dot) at all latitudes. If the full pumping is also included, the polar toroidal fields are efficiently advected inwards, and the toroidal magnetic activity survives only at the observed latitudes near the equator. With regard to the parity of the magnetic field, only models that combine turbulent pumping with near-surface shear always converge to the dipolar parity. Conclusions. This result suggests that, under the Babcock-Leighton approach, the equartorward motion of the observed magnetic activity is governed by the latitudinal pumping of the toroidal magnetic field rather than by a large scale coherent meridional flow. Our results support the idea that the parity problem is related to the quadrupolar imprint of the meridional flow on the poloidal component of the magnetic field and the turbulent pumping positively contributes to wash out this imprint.

Chromic materials for responsive surface-enhanced resonance Raman scattering systems: a nanometric pH sensor

The use of chromic materials for responsive surface-enhanced resonance Raman scattering (SERRS) based nanosensors is reported. The potential of nano-chromic SERRS is demonstrated with the use of the halochrome methyl yellow to fabricate an ultrasensitive pH optical sensor. Some of the challenges of the incorporation of chromic materials with metal nanostructures are addressed through the use of computational calculations and a comparison to measured SERRS and surface-enhanced Raman scattering (SERS) spectra is presented. A strong correlation between the measured SERRS and the medium's proton concentration is demonstrated for the pH range 2-6. The high sensitivity achieved by the use of resonance Raman conditions is shown through responsive SERRS measurements from only femtolitres of volume and with the concentration of the reporting molecules approaching the single molecule regime.

Intracellular Ca(2+) Regulation During Neuronal Differentiation of Murine Embryonal Carcinoma and Mesenchymal Stem Cells

Changes in intracellular Ca(2+) concentration ([Ca(2+)](i)) play a central role in neuronal differentiation. However, Ca(2+) signaling in this process remains poorly understood and it is unknown whether embryonic and adult stem cells share the same signaling pathways. To clarify this issue, neuronal differentiation was analyzed in two cell lines: embryonic P19 carcinoma stem cells (CSCs) and adult murine bone-marrow mesenchymal stem cells (MSC). We studied Ca(2+) release from the endoplasmic reticulum via intracellular ryanodine-sensitive (RyR) and IP(3)-sensitive (IP(3)R) receptors. We observed that caffeine, a RyR agonist, induced a [Ca(2+)](i) response that increased throughout neuronal differentiation. We also demonstrated a functional coupling between RyRs and L-but not with N-, P-, or Q-type Ca(v)1 Ca(2+) channels, both in embryonal CSC and adult MSC. We also found that agonists of L-type channels and of RyRs increase neurogenesis and neuronal differentiation, while antagonists of these channels have the opposite effect. Thus, our data demonstrate that in both cell lines RyRs control internal Ca(2+) release following voltage-dependent Ca(2+) entry via L-type Ca(2+) channels. This study shows that both in embryonal CSC and adult MSC [Ca(2+)](i) is controlled by a common pathway, indicating that coupling of L-type Ca(2+) channels and RyRs may be a conserved mechanism necessary for neuronal differentiation.

Programmable diagnostic devices made from paper and tape

This paper describes three-dimensional microfluidic paper-based analytical devices (3-D mu PADs) that can be programmed (postfabrication) by the user to generate multiple patterns of flow through them. These devices are programmed by pressing single-use 'on' buttons, using a stylus or a ballpoint pen. Pressing a button closes a small space (gap) between two vertically aligned microfluidic channels, and allows fluids to wick from one channel to the other. These devices are simple to fabricate, and are made entirely out of paper and double-sided adhesive tape. Programmable devices expand the capabilities of mu PADs and provide a simple method for controlling the movement of fluids in paper-based channels. They are the conceptual equivalent of field-programmable gate arrays (FPGAs) widely used in electronics.

A critical reassessment of elastic unloading in sharp instrumented indentation experiments and the extraction of mechanical properties

This work examines the extraction of mechanical properties from instrumented indentation P-h(s) curves via extensive three-dimensional finite element analyses for pyramidal tips in a wide range of solids under frictional and frictionless contact conditions. Since the topography of the imprint changes with the level of pile-up or sink-in, a relationship is identified between correction factor beta in the elastic equation for the unloading indentation stage and the amount of surface deformation effects. It is shown that the presumption of a constant beta significantly affects mechanical property extractions. Consequently, a new best-fit function is found for the correlation between penetration depth ratios h(e)/h(max), h(r)/h(max) and n, circumventing the need for the assumption of a constant value for beta, made in our prior investigation [Acta Mater. 53 (2005) pp. 3545-3561]. Simulations under frictional contact conditions provide sensible boundaries for the influence of friction on both h(e)/h(max) and h(r)/h(max). Friction is essentially found to induce an overestimation in the inferred n. Instrumented indentation experiments are also performed in three archetypal metallic materials exhibiting distinctly different contact responses. Mechanical property extractions are finally demonstrated in each of these materials.

Multi-actuated functionally graded piezoelectric micro-tools design: A multiphysics topology optimization approach

Micro-tools offer significant promise in a wide range of applications Such as cell Manipulation, microsurgery, and micro/nanotechnology processes. Such special micro-tools consist of multi-flexible structures actuated by two or more piezoceramic devices that must generate output displacements and forces lit different specified points of the domain and at different directions. The micro-tool Structure acts as a mechanical transformer by amplifying and changing the direction of the piezoceramics Output displacements. The design of these micro-tools involves minimization of the coupling among movements generated by various piezoceramics. To obtain enhanced micro-tool performance, the concept of multifunctional and functionally graded materials is extended by, tailoring elastic and piezoelectric properties Of the piezoceramics while simultaneously optimizing the multi-flexible structural configuration using multiphysics topology optimization. The design process considers the influence of piezoceramic property gradation and also its polarization sign. The method is implemented considering continuum material distribution with special interpolation of fictitious densities in the design domain. As examples, designs of a single piezoactuator, an XY nano-positioner actuated by two graded piezoceramics, and a micro-gripper actuated by three graded piezoceramics are considered. The results show that material gradation plays an important role to improve actuator performance, which may also lead to optimal displacements and coupling ratios with reduced amount of piezoelectric material. The present examples are limited to two-dimensional models because many of the applications for Such micro-tools are planar devices. Copyright (c) 2008 John Wiley & Sons, Ltd.

Anisotropic Etching and Nanoribbon Formation in Single-Layer Graphene

We demonstrate anisotropic etching of single-layer graphene by thermally activated nickel nanoparticles. Using this technique, we obtain sub-10-nm nanoribbons and other graphene nanostructures with edges aligned along a single crystallographic direction. We observe a new catalytic channeling behavior, whereby etched cuts do not intersect, resulting in continuously connected geometries. Raman spectroscopy and electronic measurements show that the quality of the graphene is resilient under the etching conditions, indicating that this method may serve as a powerful technique to produce graphene nanocircuits with well-defined crystallographic edges.