1731 research outputs found
Sort by
Regulation on Safety and civil Liability of intelligent autonomous Robots: the case of smart Cars
Nowadays science and technology offer us artificial intelligence (AI) “embodied” in robots. They are able to self-learn, self-organize and self-reproduce, thanks to genetic algorithms, artificial neural networks and other tools. The focus of this research includes results from the diffusion of a social phenomenon consisting in the application of robots in the most disparate realities (industrial and domestic). Robotics is the AI branch whose aim is to build machines that are able “to feel, to think and to act”. A robot is a complex system that integrates many AI results: such as vision, natural language, study of the movement, communication, machine learning, and knowledge representation and planning. Robotics presents a fascinating and unexpected scenario. Everything we have seen until now - cars, computers, mobile phones and internet - is an AI product and robots in a near future will be used in factories, in yards, in offices, and they will work as nurses in hospitals and in households. The development of robots raises new ethical, legal and social issues, such as the allocation of civil liability when a robot harms a human being. If software agents can process and take decisions autonomously who will be held liable in case of harmful consequences of these decisions? The manufacturer, the programmer, the owner, or other subjects? The answer requires an analysis on both ex ante and ex post tortious event.
* * *
Aujourd’hui science et technologie nous offrent l’intelligence artificielle (AI) «incarnés» dans les robots. Ils sont capables d’apprendre, de s’organiser et de se reproduire, grâce à des algorithmes génétiques, réseaux de neurones artificiels et d’autres outils. L’objectif de cette recherche comprend les résultats de la diffusion d’un phénomène social consistant dans l’application de robots dans les réalités les plus divers (domestiques et industriels). Robotique est le secteur du AI dont le but est de construire des machines qui sont en mesure de «sentir, penser et agir». Un robot est un système complexe qui intègre de nombreux résultats du AI: tels que la vision, la langue naturelle, l’étude du mouvement, la communication, l’apprentissage machine, et la représentation de la connaissance et de la planification. Robotique présente un scénario fascinant et inattendu. Tout ce que nous avons vu jusqu’à présent - les voitures, les ordinateurs, les téléphones mobiles et Internet - est un produit du AI et les robots dans un proche avenir seront utilisés dans les usines, les gares, les bureaux et ils vont travailler comme infirmières dans les hôpitaux et dans les ménages. Le développement de robots soulève de nouvelles questions éthiques, juridiques et sociales, telles que l’attribution de la responsabilité civile quand un robot nuit un être humain. Si des agents logiciels peuvent traiter et prendre des décisions de manière autonome qui sera tenu responsable en cas de conséquences néfastes de ces décisions? Le fabricant, le programmateur, le propriétaire, ou d’autres sujets? La réponse exige une analyse relative ex ante et ex post le fait générateur
Investigating Communication and Social Behaviour Using Wearable Sensor Technology
The behaviour that we exhibit contributes to the message that is communicated to those that we are interacting with and can have an impact on how the message is conveyed and interpreted. Nonverbal behaviour is just as important to be aware of as well as what is being said, as the subtleties of behaviour can impact the outcome of interactions. Advancements in research technologies have allowed us the chance to investigate natural human behaviour is a variety of settings outside of the laboratory, however, some gaps in the understanding of behaviour exist. The aim of this thesis is to investigate communication and social interactions in a variety of settings, paying particular attention to the methods of data collection, specifically the use of wearable sensors, to investigate phenomenon from social psychology. The thesis aims to address three specific research questions; 1) if can we predict stress using a combination of nonverbal behavioural cues along with physiological measurements, 2) understand the factors affecting happiness and productivity in the workplace from features of communication taken from wearable sensors and 3) determine the stressors that can be characterised from communication patterns assessed through Call Detail Records and smartphone sensors. The studies presented here focus on the nonverbal aspects of communication that can be measured through wearable and sensing devices. In the three types of scenarios that are detailed in the different chapters, the interactions considered are face to face meetings in a one on one interaction, co-location within a defined space in an organisation and the communications of a widely dispersed community. The interactions are recorded by wearable devices such as the Affectiva Q sensor, the Sociometric Badge, and smartphones equipped with sensing capabilities in the form of the funf and P-OWL platforms for data recording, among other forms of data collection. Each of the studies included aspects of self reported assessments that were used as a ground truth measurement of affect: these were annotations of stress, self reports of fear of negative evaluation, self perception, positive and negative affect and stress, among others. The goal was to examine how to use digital traces of behavioural expressions to have a greater understanding of these interactions and how the way in which we interact with others has an impact on the individual. The work from this thesis adds to the existing literature on these various issues by addressing the research questions from a novel perspective. The studies found support for each of the research questions and by using a mixed methods approach and digital traces from wearable sensors gained insights into how communication impacted the individual, revealing the important aspects of communication and their effect on stress, productivity and well-being
Social interaction analysis in in videos, from wide to close perspective
In today’s digital age, the enhancement of the hardware technology has set new horizons on the computer science universe, asking new questions, proposing new solutions and re-opening some branches that have been temporary closed due to the overwhelming computational complexity. In this sense many algorithms have been proposed but they have never been successfully applied in practice up to now.
In this work we will tackle the issues related to the detection and the localization of an interaction conducted by humans. We will begin analysing group interactions then moving to dyadic interactions and then elevate our considerations to the real world scenario. We will propose new challenging datasets, introducing new important tasks and suggesting some possible solutions
Middle Managers training in Italy: a firm-level analysis
This research presents for the first time panel evidence on the performance and wage effects of middle mangers' training in Italy. It also offers an analysis of the determinants of training by investigating the relationship between training provision and firm characteristics. It is based on a rich and reliable panel dataset covering Italian firms for the years 2006-2011. Several estimation techniques and model specifications are implemented to argue that middle mangers' training significantly increases productivity and to prove the existence of a 'too-much-of a good' thing effect. Similarly a simulation of the wage effects of training shows how the positive magnitude of the coefficients seems to be severely affected by the training measure used.
All the analysis implemented suggests the importance of firm's size and geographic location in explaining training provision, firm's performance, and wages. Indeed larger firms and firms located in Northern Italy appear to be more likely to offer training of higher intensity, quality and variety. Furthermore training is found to be more effective for what concerns business results and individual wages. This might suggest that medium-small firms and firms located in Southern areas could be trapped in a circle of scarce, low quality and standardised training provision with obvious implications on its efficiency and returns
Seismic Behavior of concrete filled steel Tubular Built-up columns
With the advantages of CFST built-up columns, including the higher confinement in the concrete, delay of the steel local buckling, higher compressive and flexural strength, earthquake and fire resistance, rapid construction, savings in the construction costs, etc. CFST built-up columns are increasing adopted in structural members with larger load eccentricity ratio and slenderness ratio, such as stadium, industrial buildings, bridge pier and pillar, and electrical transmission tower. However, the research is mainly focused on static performance, seldom research has been reported on the dynamic behavior of CFST built-up columns. The present research investigates the seismic behavior of CFST built-up columns. A detailed literature survey on the CFST built-up structures, including mechanical characteristics, applications, ductility in seismic design, previous experimental researches, and finite element formulation, is firstly illustrated. Six specimens with different grades of concrete and brace arrangements are designed and tested subjected to cyclic loading. The hysteretic behavior, such as failure mode, deformed shape, displacement ductility, rigidity and strength degradation, and energy dissipation capacity of test specimens are discussed. The corresponding validated finite element model (FEM) simulations are developed for parametric analysis, to discuss the hysteretic behavior, affected by axial load ratio, chord spacing, brace spacing, diameter to thickness ratio, and steel yield strength. Results indicate that the hysteretic characteristics of specimens are saturated and exhibited good ductility. The concrete strength and steel yield strength played a slight effect to the displacement ductility factor. While the ductility will be significantly affected by axial load ratio and geometrical types. Based on extended parametric analysis and regression analysis, a simplified method, consisted by equivalent slenderness ratio, axial load ratio and steel yield strength, is proposed to calculate the displacement ductility factor of CFST battened columns and laced columns, respectively. The accuracy is validated with test results. After that, to investigate the seismic performance of built-up columns used in practice, an innovative lightweight bridge with CFST composite truss girder and CFST lattice pier is studied as case study. For the purpose, FEM simulation and shaking table test are carried out. The FEM results agree with experimental data. In addition, the plastic hinges were predicted under transverse and longitudinal excitation respectively, revealed that CFST built-up columns has a favorable seismic performance
On Boolean functions, symmetric cryptography and algebraic coding theory
In the first part of this thesis we report results about some “linear” trapdoors that can be embedded in a block cipher. In particular we are interested in any block cipher which has invertible S-boxes and that acts as a permutation on the message space, once the key is chosen. The message space is a vector space and we can endow it with alternative operations (hidden sums) for which the structure of vector space is preserved. Each of this operation is related to a different copy of the affine group. So, our block cipher could be affine with respect to one of these hidden sums.
We show conditions on the S-box able to prevent a type of trapdoors based on hidden sums, in particular we introduce the notion of Anti-Crooked function. Moreover we shows some properties of the translation groups related to these hidden sums, characterizing those that are generated by affine permutations. In that case we prove that hidden sum trapdoors are practical and we can perform a global reconstruction attack. We also analyze the role of the mixing layer obtaining results suggesting the possibility to have undetectable hidden sum trapdoors using MDS mixing layers. In the second part we take into account the index coding with side information (ICSI) problem. Firstly we
investigate the optimal length of a linear index code, that is equal to the min-rank of the hypergraph related to the instance of the ICSI problem. In particular we extend the the so-called Sandwich Property from graphs to hypergraphs and also we give an upper bound on the min-rank of an hypergraph taking advantage of incidence structures such as 2-designs and projective planes. Then we consider the more general case when the side information are coded, the index coding with coded side information (ICCSI) problem. We extend some results on the error correction index codes to the ICCSI problem case and a syndrome decoding algorithm is also given
A nonlinear viscoplastic double yield surface constitutive model for geologic materials
A new constitutive model is proposed to describe the stress-strain-time behaviour of clays and organic soils. This approach is based on the framework of overstress viscoplastic theory and can be applied to any elastoplastic (inviscid) model.The most innovative idea of the proposed model is the identification of two components of strain for each deformation mechanism. In particular, in both the elastic and the plastic regimes a fast and a slow strain mechanism are assumed to be present. To this aim the constitutive model is based on two yield surfaces based on overstress viscoplasticity theory: one is quasi-instantaneous (for the fast part of plastic deformations) and the other is viscous (for the slow part of plastic deformations). This assumption permits the reliable simulation of the mechanical behaviour of a wide range of clayey soils: from inorganic clay (a.e. kaolinite and bentonite), to organic clays and peats. A further interesting aspect of the proposed model is the capability of simulate both the normally consolidated and the overconsolidated regime within a unique constitutive approach. An extensive experimental program on three different peaty soils has been performed to validate the constitutive model. The experimental tests include a wide range of oedometric tests (with very different loading conditions), and a series of triaxial tests conducted on NC and OC peats. The model has been validated also by simulating the settlements of a real embankment founded on a thick layer of organic clay
A torsion pendulum ground test of the LISA Pathfinder Free-fall mode
The LISA Pathfinder geodesic explorer mission for gravitational wave astronomy aims to measure a residual differential acceleration noise approaching the femto-m/s^2/Hz^1/2 levels needed for eLISA. This measurement is complicated by a large, roughly nm/s^2, stable
differential gravitational acceleration that must be actively compensated in order to hold the test particles centred inside an orbiting apparatus. The actuation force applied to compensate this effect introduces a dominant source of force noise in the mission noise budget. To suppress this noise source and avoid actuation instabilities, a “free-fall” actuation control scheme has been designed: actuation is limited to brief impulses, with test masses in free fall in between two “kicks”, with this actuation-free motion then analysed for the remaining
sources of acceleration ultra noise. In this work, we present the results from an extensive on-ground torsion
pendulum test of this free-fall technique, which to date allows an equivalent acceleration noise measurement at the 100 fm/s^2/sqrt(Hz), a factor 3 above the LISA Pathfinder spec. We will discuss both experimental and analysis limitations to the ground experiment and some
implications for the flight test
The Political Economy of Agricultural Cooperatives in Bosnia and Herzegovina: Towards a Sustainable Rural Development Model
Agricultural cooperatives emerge as tools through which individual farmers meet their economic and social needs and they do so in a manner that allows them both to economize on costs and to disperse the risks associated with pursuing these needs individually. They are particularly useful for enhancing economic development of transition countries in which basic market economy infrastructure is either underdeveloped or is altogether missing. However, agricultural cooperatives do not always deliver the desired level of rural development. Although the literature sometimes takes this to mean that the model itself is defective, I argue to the contrary. In this thesis, I propose that it is the evolutionary path of cooperative idea, its implementation in reality and the way in which it interacts with its institutional surroundings that condition the ability of the model to perform. In other words, cooperative idea is not immune to its political and economic context but rather it is molded by it, and sometimes to the point that it no longer resembles its original substance.
In line with that, the main objective of this thesis is to look into factors that either stimulate or discourage development and functioning of agricultural cooperatives in a context of post-socialist and post-conflict Bosnia and Herzegovina (BiH). In doing so, the thesis adopts a new institutionalist frame of analysis combining it with the insights from the economics of the third sector and the cooperative theory to highlight both the evolutionary nature of cooperative idea as well as its embeddedness in the socio-economic context. The research relied on quantitative and qualitative approaches and gathered data from field work and secondary sources.
The main findings can be summarized in the following several points: both formal and informal institutions have shaped the way in which cooperatives are understood and utilized by farmers in BiH; cooperatives in post-socialist and post-conflict settings require legal clarity that not only sets them apart from other types of firms in the market but also from degenerated forms of cooperatives that exist to serve interests of few individuals rather than cooperative members and their communities; when judged by the standards that apply to cooperatives in economically advanced societies, it is safe to state that there are very few true cooperatives in BiH; given the structure of agricultural market and number of farmers, there is a lot of potential in utilizing the cooperative model for purposes of rural development. However, using cooperatives for development purposes requires a basic alignment between the features of institutional environment and cooperative organizational characteristics. If stimulated properly through positive policy changes, cooperatives can exhibit transformative potential that is best reflected in how they empower their patrons as well as contribute to the development of their communities
Verbs and nouns in awake neurosurgery: needs and answers
Awake surgery is a surgical procedure applied to people with brain tumors. Patients are kept awake during surgery to perform language tasks during electrical stimulation of cortico-subcortical brain structures. When the application of electrical pulses to the brain repeatedly triggers errors, this is taken as an indication that the area is critically involved in the language skills probed by the task, and therefore should not be removed, as this would result in postoperative language deficits. In this thesis, we focused on the role of production tasks using finite verbs and the evaluation of their ability to assess language processing in comparison with the current gold standard (i.e., object naming). As a preliminary step in the development of novel, controlled object and verb production tasks, we critically reviewed the literature. We discussed published studies that proposed intraoperative tasks requiring the production of verbs in isolation and in sentences, and we graphically represented the stimulation sites and error types that were reported in each study. Subsequently, we discussed the advantages and disadvantages of each from a cognitive neuroscience and neurolinguistic perspective. With this information in mind, we constructed and validated two naming tasks for native speakers of Italian undergoing awake surgery – an object naming task and a finite verb production task. We administered the finite verb production task and other picture naming tasks to subjects with post-stroke aphasia, and established that a task that uses finite verbs shows a strong correlation with communicative abilities in daily life. Also, we introduced and compared our standardized finite verb task in the operating theater and compared it with an object naming task. Our tests are currently and successfully used in different surgery wards in Italy and overseas. This work can be extended to other neurological populations such as stroke, dementias, and epilepsy