Search

Today the smart devices are more and more con- nected to each other. The people interact with these devices to execute simple and basic operations. The increase of the personal assistants has simplified these interactions thanks to the vocal commands. However they are not smart enough to adapt to the user lifestyle, forcing the user to manually configure the actions to be performed when a specific event is triggered (top-down approach). Therefore the interactions between the devices are conditioned by the user perception of its lifestyle, risking to miss some scenarios that the human does not sense. In this work we propose a proactive approach (bottom-up) to the interaction between the personal assistants and other IoT devices, that enables to modify at run-time the actions and to anticipate the user requests, self-configuring to provide more services. This approach uses a main personal assistant that analyse the user interactions to the IoT devices and other personal assistants, writes and installs the driver code to the interested devices. The interaction between the IoT devices and the main personal assistant is supported by WoX.

The proposed paper discusses different technologies suitable for non-invasive road infrastructure inspection. Specifically, the study is focused on the description of Unmanned Aerial Vehicles (UAVs) and cars equipped with infrared thermal cameras and Ground Penetrating Radar (GPR) antennas able to detect the health status of roads and bridges. The analysis enhances the methodologies to extract information from radiometric images and radargrams, such as infiltrations and asphalt defects, thus defining new inspection procedures and approaches. Finally, the paper proposes a full equipped car detection system, embedding other sensors enabling data fusion. The analysed system is suitable for an automatic mapping of the whole road infrastructure, and for predictive maintenance.

An effective method of managing water pipelines is essential for maximizing water efficiency and preserving water quality for both human and agricultural use. The aim of this study is to determine and test such a method for large-scale water piping networks intended for irrigation. Despite the considerable efforts being made worldwide to reduce the waste of potable water, nowadays there is still no widely used technique for this purpose. The most common methodologies for the detection of leaks use acoustic emission, gas detection, various types of sensors, such as hydraulic pressure and temperature. Unfortunately, these methods present critical issues and are difficult to apply on a large scale. This paper proposes an approach based on infrared thermography that can be associated with the use of a drone in order to reach and monitor large areas in a short time. A real time monitoring and decision support technological platform oriented towards water efficiency and the computation of hydrogeological risks has been implemented. In fact, the system integrates different detection technologies, as the Infrared Thermographic (IRT) technique is supported by the Ground Penetrating Radar (GPR) for the inspection of buried pipelines. Artificial Intelligence (AI) algorithms have been used for radiometric image processing with the aim of identifying water leaks even when these are latent. In particular, a Convolutional Neural Network (CNN) was implemented to identify leakages. As regards the calculation of hydrogeological risk, algorithms based on logical conditions have been implemented that allow to match the meteorological data and the hydrogeological risk maps of the places of interest, in order to predict the hydrogeological risk of the territory.

In this paper is proposed a methodological approach for the detection of leaks in water pipelines. The approach is based on the use of Infrared Thermography (IRT) for the real time monitoring, and of Artificial Neural Networks (ANNs) for the identification of potential leaks not easy visible by IRT. The input data source consists of radiometric data processed by Convolutional Neural Networks (CNNs) providing as output information about the presence or absence of the water leakages. A preliminary study has been carried out about a fixed monitoring station created to identify leaks in underground pipelines. In addition, has been implemented a platform to remotely acquire the thermograms and to analyze them by CNN networks detecting leakages, combined with Long Short-Term Memory (LSTM) neural networks, and image filtering algorithms, such as image segmentation and active contour snake approach. The LSTM network allows the prediction and calculation of the propagation trend of the leak plume. Finally, image filtering improves the visualization of leaks as it allows to draw the contours of the pixel clusters representing the leakages areas in the thermograms. The work was developed within the research framework of an industrial project. The proposed approach is suitable also for oil spill and gas leakages detections.

The demand of innovative solutions to afford healthy and safe lifestyle, for elderly people, are increasing lead by the growth of the ageing population and age-related diseases. Moreover, the availability of smart devices based on IoT technologies for personal environments has evolved the demand for tools to monitor and remotely control the home equipment. A new generation of remote control is available: Electronic Personal Assistants. These devices could be controlled using natural user interfaces, help user to simplify the interaction with home automation and other smart devices. The personal assistants use machine learning strategies to interact with the user. The solutions available in the market are often expensive and, moreover, are limited and legacy since based on proprietary hardware and software. This paper aims to present Alfred, a flexible and modular smart personal assistant based on low-cost hardware and open source machine learning software, in the context of the City4Age, an H2020 Project, to provide Elderly-friendly ambient assisted living.

The infrared thermography approach is applied for the monitoring of fresh foods located into a refrigerator by controlling the temperature and automatizing the deterioration risk alert. Specifically, the paper discusses the results about infrared thermography methodology combined with image processing technique. The study is performed to check during the time the evolution of the meat temperature when placed out of the refrigerator at room temperature. The experimentation is performed by analyzing the meat oxidation time evolution and checking the minimum temperature behavior defining food quality. A K-Means algorithm is applied in order to locally process the radiometric image highlighting the oxidation effect. Finally are discussed different aspects about radiometric approaches.

The proposed work focused on a methodological approach to check defects by adopting the Watershed approach. The adopted image segmentation technique was applied by estimating the defect extent of manufactured pieces. In particular, the experimental tests were performed by characterizing defected holes and defining tolerance gaps on the performed measurements. Information about image segmentation sensitivity was provided by varying camera distances from the sample. Additionally, the defect identification was proved by applying the K-Means algorithm as a processing tool on the analyzed image. The work was developed within a research industrial project.

The proposed work describes an architecture able to optimize production supporting wheat transformation processes, developed within the framework of an industry research project. The architecture is designed in order to integrate and enable different systems and technologies including big data systems and artificial intelligence algorithms. Preliminary results about detected field data and data flow implementation are discussed. A particular attention is focused on predictive maintenance of production machines by infrared thermography imaging and accelerometer signal processing.

The present paper proposes a methodological approach to evaluate welding quality in the context of tank production. In particular, infrared thermography was adopted to control the structural homogeneity achieved after welding. The analysis was implemented by applying the K-Means algorithm, morphological image processing and artificial neural network based on the Long Short Term Memory - LSTM - technique. The adopted approach was chosen to perform different image post-processing analysis and therefore highlights, identifies, and quantifies welding inhomogeneities, as cracks and defects. The work was developed within the research framework of an industrial project. The proposed approach could be implemented in inline production systems which integrate an artificial intelligence processor for real time quality monitoring.

A radio-frequency planar antenna 1 for applications in inductively-coupled plasma processing is investigated at atmospheric pressure within the ANSYS/FLUENT environment 2. Continuity, momentum and energy equations are solved for optically thin argon plasma in LTE conditions and laminar flow, coupled with electromagnetic field equations, solved by means of an User Defined Scalar approach in three different domains 3 dielectric, plasma and legs of the antenna).

High neural selectivity is an important indicator of how well a neural electrode will perform in a clinical setup. Among other factors, neural selectivity of an electrode strongly depends on its implant location. Today, the neural selectivity is evaluated by performing animal experiments and consequent offline analysis. The most selective electrode implant location is therefore difficult to predict. Here we developed a software tool to characterize neural electrodes' selectivity in real-time, in vivo. A graphical user interface (GUI) allows controlling the software tool. Our tool is adapted to use a commercially available neurostimulator (TDT, Tucker-Davis Technologies, US) but can be easily adapted to other systems. Electromyographic ( EMG) recruitment curves and selectivity indexes (SIs) were obtained stimulating the sciatic nerve of four pigs. The functionality, execution time of stimulation and correct SI calculation by the software-platform were evaluated by testing twelve intraneural electrodes.