Abstract: Anomaly detection is the identification of any event that falls outside what is considered ‘acceptable behaviour’. This work investigates anomaly detection for automated visual inspection in the context of industry automation (‘Industry 4.0’). For this task we propose a machine vision procedure based on visual feature extraction and one-class k nearest neighbours classification. The method requires only samples of normal (non-defective) instances for the training step. We benchmarked our approach using seven traditional (‘hand-designed’) colour texture descriptors and five pre-trained convolutional neural networks (CNN) ‘off-the-shelf’. Experimenting on nine image datasets from seven classes of materials (carpet, concrete, fabric, layered fused filament, leather, paper and wood), each containing normal and abnormal samples, we found overall accuracy in the range 82.0%–90.2%. Convolutional networks off-the-shelf performed generally better than the traditional methods, although – interestingly – this was not true for all the datasets considered. No visual descriptor clearly emerged as the all-purpose best option.

Keywords: Anomaly detection | Colour | Convolutional neural networks | Texture | Visual descriptors

Abstract: Snap-fit joints represent a simple, economical and straightforward way of joining two different components. The design of the snap-fit joint is usually performed evaluating peak stresses that must be tolerated by the material without incurring into failure or plastic deformations; in addition, the force needed to join and disassemble parts is estimated in relation to ergonomic issues. Finally, the retention force, that is the force required to start disjoining parts, needs to be estimated. The evaluation of peak stresses or insertion/retention/removal forces is commonly performed through finite element method, having identified the respective deformed configuration. A different approach has been here followed considering that it is not trivial to identify the most critical condition in a full joining/disjoining cycle, when complex geometries are being considered. In detail, the snap joint has been modelled as a multibody model including a flexible body, which replicates the part that undergoes major deflections during the process. The model has been validated against experimental force – time curves, recorded for an existing joint, and it has been used to optimize a parametrised snap-fit design. As a result, the joining force has been reduced up to −84%; the disassembly force has been reduced up to −86% and the retention force has been incremented up to +7%. On the whole, a numerical framework to study these joints has been established, keeping the computational time reasonably low (about 40 min for the entire insertion and removal simulation).

Keywords: Geometrical modelling | Multibody model | Plastic components design | Snap-fit joint | Tolerance analysis

Abstract: In this work a new approach for the creation of Articulated Total Body (ATB) models for person-specific multi-body simulations is presented, with the main aim of overcoming limitations related to classical multi-ellipsoids ATB models, based on regression equations having only the weight and the height of the subject as input. The new methodology is based on a Statistical Shape Model (SSM), morphable according to up to 24 input parameters: the SSM was obtained from Principal Component Analysis (PCA), applied on a wide database of 3D human scans (CAESAR). The so obtained geometry can be segmented automatically to generate body segments with the respective inertial properties (mass, principal moments of inertia, and centres of mass location). The routine has been tested on a random set of 20 male subjects and the classical multi-ellipsoids models were compared to these in terms of inertial properties and 3D external geometry: the highest differences were registered at the abdomen and the thighs for what concerns the mass (60%), principal moments (75%) and centres of mass (50 mm) properties; the trunk, the shoulder and the calves are the most critical areas for the external geometry (average distance between the anthropomorphic and ellipsoids models equal to 50 mm). A contribution has been made to build person-specific multibody models. This is a valuable method since approximations made by multi-ellipsoidal models have resulted to be relevant at specific body areas, and personalised models can be a support to design and to forensic analyses.

Keywords: 3D parametric human model | Articulated total body | Forensic biomechanics | Multibody analysis | Principal component analysis (PCA)

Abstract: The aim of this research is to develop patient-specific 3D mandible models, based on a limited number of measurements taken on the patient. Twenty Computed Tomography scans were used to build the respective 3D cad models of the mandible. Fifteen of these models were given as an input to a Principal Component Analysis software, and eight ‘principal’ mandible morphologies were produced. The following step was to identify the most efficient landmarks to ‘weight’ these morphologies when building a patient-specific model. Two further mandible computed tomography scans (a ‘normal’ mandible and a ‘severely resorbed’ one) were used to test the full procedure and to assess its accuracy. The accuracy of the 3D morphed surface resulted to range between 0.025 and 3.235 mm for the ‘normal’ mandible and between 0.012 and 1.149 mm for the ‘severely resorbed’ one having used eight landmarks to morph a ‘standard’ mandible. This work demonstrates how patient-specific models can be obtained registering the position of a limited number of points (on panoramic x-ray or on the physical model), reaching a good accuracy. This allows performing patient-specific planning and numerical simulations even for those cases where a computed tomography scan would not be available. In fact, this procedure can be interfaced with mesh morphing algorithms to automatically build finite element models. The accuracy of the procedure can be further improved, widening the mandibles computed tomography scans database and optimizing landmarks position.

Keywords: Morphing | Patient-specific models | Principal Component Analysis

Abstract: The final subject position is often the only evidence in the case of the fall of a human being from a given height. Foreseeing the body trajectory and the respective driving force may not be trivial due to the possibility of rotations and to an unknown initial position and momentum of the subject. This article illustrates how multibody models can be used for this aim, with specific reference to an actual case, where a worker fell into a stair well, prior to stair mounting, and he was found in an unexpected posture. The aim of the analysis was establishing if this worker was dead in that same place, if he had been pushed, and which was his initial position. A multibody model of the subject has been built (“numerical android”), given his stature and his known mass. Multiple simulations have been performed, following a design of experiments where various initial positions and velocity as well as pushing forces have been considered, while the objective function to be minimized was the deviation of the numerical android position from the actual worker position. At the end of the analysis, it was possible to point how a very limited set of conditions, all including the application of an external pushing force (or initial speed), could produce the given final posture with an error on the distance function equal to 0.39 m. The full analysis gives a demonstration of the potentiality of multibody models as a tool for the analysis of falls in forensic inquiries.

Keywords: accident | android | biomechanics | crime | doe | fall | forensic | multibody

Abstract: This work is focused on the analysis of the fall of a human being from a given height. With reference to forensic disputes, the final subject position is often the only evidence and foreseeing the body trajectory and the respective driving force may not be trivial. This article illustrates how multibody models can be used for this aim. A multibody model of a human subject has been built, given his stature and his known mass. This model was made of 15 segments, whose inertial properties, joint centres and volumes were deduced from anthropometric databases. This model was validated against experimental tests performed on a Hybrid III dummy: it was able to reproduce the peak impact head force with an error lower than about 10%. Some examples are produced to illustrate the usefulness of this validated model as a tool for the analysis of falls, and how it can be easily parametrized to make multiple simulations with different initial conditions/environment configurations. As such it is a valuable tool for forensic analyses.

Keywords: Anthropometric data | Fall from height | Forensic biomechancis | Multibody model

Abstract: An obstacle to the diffusion of additive technology is the difficulty of predicting the residual stresses introduced during the fabrication process. This problem has a considerable practical interest as evidenced by the abundant literature on residual stresses and distortion induced by the SLM (Selective Laser Melting) and EBAM (Electron Beam Additive Manufacturing). The purpose of this paper is to evaluate the effect of different process parameters on the heat distribution and residual stresses in components made with SLM technique. Three aspects are developed and illustrated: a) thermomechanical modeling of the growth process, based on Finite Elements (FE), which considers changes in the behavior of the material (powder→liquid→solid) through the finite element "birth" and "death" technique that enables the progressive activation of the elements as the component grows; b) sensitivity analysis of the model to the physical characteristics of the material (conductivity, specific heat capacity, Young's modulus). This is an important aspect allowing to focus on the most significant parameters to be determined experimentally with high reliability; c) evaluation of the effects of different process parameters (laser power, scan speed, overlap between adjacent paths) on the process. The article illustrates the theoretical thermal model and the detail of the strategy used in the FE analysis. The most influential characteristics of the material are highlighted and, finally, general criteria for choosing the optimal combination of process parameters to limit the residual stresses are provided.

Keywords: Additive Manufacturing | FE model | Residual stresses | Selective Laser Melting

Abstract: Nowadays Additive Manufacturing (AM) is going through a very fast development, spreading in many different mechanical contexts. The main advantages of this technology are: production costs reduction (prototype realization time reduction, raw material consumption reduction, almost zero manpower needed...), significant reliability (compared to the standard production process) and last but not least extreme freedom in product shape design. The last characteristic makes it possible to adopt new design approach focusing on component shape and material distribution optimization; a new design paradigm must be developed to fully take advantage of these opportunities: The designer can develop new concepts with very complex shapes and sophisticated topological solution owing to opportunities yielded by AM with in mind only the week limitations given by this technology. In detail this work aims to highlight a new design strategy that consist of a combination of structural optimization tools (Topology Optimization TO) and non-contact stress field measurement technique (based on thermo-elasticity). The goal is to develop an iterative design procedures which links the design shape optimization with the experimental stress evaluation, allowing a wise material distribution in order to enhance the resistance. The idea is to accomplish an initial designing phase, letting the designer free to define a first rough design concept taking into account the information provided by the TO to exploit the material in the best way. Then, the concept must be verified in both: model numerical F.E.M. analysis and prototype experimental evaluation of the stress field. Eventually, according to the verification analysis results, the model will be modified to reach the desired requirements in terms of allowed deformation, stress resistance and fatigue life. The paper will display the optimization technique iterative process (based on Solid Isotropic Material with Penalization - SIMP - scheme) in a general way and through a practical example. As a reference, this methodology has been applied to a specific test case in order to design and optimize a new concept of a structural mechanical component of a mountain bike. The component was, first realized as a prototype in thermoplastic material and finally designed to be realized in metal for in field application.

Abstract: The new trends in the design of race catamarans has introduced new innovations as huge vertical wings instead of main soft sails, and foiling underwater wings. The scope of the paper is the development of a mathematical model of these new features to be implemented in a simulator designed for training purpose. The paper describes the principal assumptions, the simplifications, and the modeling strategies that were adopted in order to obtain a real time simulation. The mathematical model is implemented in an already existing SIMULINK simulator developed at the University of Southampton and exploits its graphical interfaces. The main new feature is the “flying” simulation. The simulator is designed to interact with the in-training team and to feedback the crewmen with realistic cues. Beside training purpose, the simulator could also be a useful tool in comparing different race strategies in order to select the most promising one.

Keywords: America’s cup | Crew training | Human interface | Real-time | Sailing simulator | Simulator

Abstract: To-day America's Cup catamarans have many innovative features such as hydrodynamic foils and rigid wings instead of soft sails. They are designed not only to float but also to heave and "fly" over the sea surface. These new features require new skills that the crew must acquire. The work presented in the paper deals with this problem and describes a foiling catamaran simulator designed for training purpose. The simulator is designed primarily to interact with the in-training team and to feedback the crewmen with realistic physical reactions in an immersive scenario; secondly the simulator gives the opportunity to compare different race strategies and to select the most promising one. The main features of the simulator are illustrated, some graphical evaluations are displayed and results are discussed.

Keywords: America's Cup | Crew Training | Human Interface | Real-time Simulator | Sailing Simulator

Abstract: Considerable progress has been made in the development of Velocity Prediction Programs (VPPs) suitable for analysis of racing yacht performance. In addition, investigations on yacht dynamics (i.e. optimal tacking procedure) are now available. While these tools will no doubt be further refined and computations speeded up, there is also a need to assess the performance of the yacht's helmsman and crew. The scope of the present study is the prediction of the performances of a yacht-crew system as a whole, by deriving numerical models for human behaviour alongside with those referred to the physics of yacht motion. The latter issue, the mechanical side of the problem, is analysed by solving yacht equations of motions in the time domain; crew inputs in terms of yacht steering and sail trim are considered. The yacht-crew system can sail a racecourse in an arbitrary wind pattern, according to strategic rules and given decision making schemata. © 2008: Royal Institution of Naval Architects.

Abstract: Reverse Engineering aims at constructing CAD models of existing objects starting from measurement data. In the classical approach a reverse engineering procedure completes in three main steps: data acquisition data processing and modeling. Herein we describe a novel semi-automatic approach for integrating CAD and reverse engineering. The basic idea consists in constructing a parametric and feature-based approximate model and matching it with measurement data to determine the unknown parameters. The approach is semi-automatic since part of the model is constructed manually by the user, while the exact values of the parameters are computed through an optimization procedure based on genetic algorithms. Two case studies are presented and critically discussed.

Keywords: genetic algorithms | geometric modeling | Reverse engineering

Abstract: This paper describes an approach for comparative evaluation of manufacturing cost in the early stage of the design process of products and assemblies. The methodological approach used here refers to Design For Manufacturing and Assembly technique (DFMA) and is addressed towards the development of optimized design solutions.

Abstract: This paper describes an approach for comparative evaluation of the response of a system in different domains in the early stage of the design process. The solution proposed here is based on the following ideas: feature based product representation, which serves as a 'shared product description'; estimation of the response of the system in a set of points (training points) through automatic simulation; categorization of the system response; generalization to other design regions through classification trees; evaluation of reliability. The result is an approximated 'model' of the system we are investigating. The level of detail of the model can be modified by changing the number of classes used to categorize the response.

Keywords: Classification trees | Cost estimation | Design-by-feature | Product design

Abstract: The challenge of racing one-design yachts is to maximize the performance of the yacht within the scope allowed by the relevant regulations. Such tuning of the yacht, for a well-policed rale, should only make possible small gains. The main area of possible performance gain is in how best an individual sailor or crew can fine tune their racing strategy. The ability to model such strategic decisions requires an understanding of both the physical behaviour of the yacht and how an individual sailor makes such decisions. The present study seeks to predict the performances of a yacht-crew system as a whole by deriving numerical models for human behaviour alongside those referring to the physics of yacht motion. The former aspect, a transposition of athletes' psychology within the racing scene, is investigated by means of questionnaires submitted to skilled athletes and structured interviews with sailing coaches. The latter issue, the mechanical side of the problem, is analysed by solving yacht equations of motions in the time domain; crew inputs in terms of yacht steering and sail trim are considered. The paper presents results from simulations in which the yacht-crew system can sail a racecourse in an arbitrary wind pattern, according to strategic and tactical rules derived by common practice and following the decision making schemata obtained above.

Abstract: A sailing simulator is described that takes into account human factors with respect to strategic decisions during a race. Models for yacht-crew interaction have been designed and implemented for the helmsman and the sail tailers, as well as a 'routing engine' that solves problems of a strategical and a tactical nature (e.g. navigation in windshifts, collision avoidance). One or more users can interact with the simulator, so that their behaviour and decision-making process can be recorded, analyzed and compared to that of the automatic crews. The software was implemented in MATLAB-Simulink and consists of two modules: 'Robo-Yacht', for solo and drag races, and 'Robo-Race' for fleet regattas. In the latter case, a real-time animation of the race is delivered to the user in real-time within a virtual reality environment. A series of case studies are presented that demonstrate realistic interactions between a yacht whose decisions are made by a sailor and a mini fleet of autonomous yachts.

Abstract: The work presented in this paper is focused on the finite element modelling (FEM) of the failure behaviour of lattice composite hollow structures that have been subjected to an external hydrostatic pressure. Furthermore, the development of an experimental procedure to measure the aforementioned resistance, and to test the FEM model is also presented. Carbon fibres composite hollow cylinders with a lattice structure and with different geometries were produced and tested. In order to develop a design tools for such structures, all the experiments performed were computer simulated using finite element modelling. The results obtained with FEM simulation provide further insight to analyze and investigate the failure mechanism. The elastic instability of the studied structures was therefore analyzed and the influence of element geometry on the collapsing resistance thereof was considered. As a result of the study it has been possible to locate three different failure modes which where strictly related to the length of the cylinders. Both the shape of the broken tube and the lever of failure stresses were correctly predicted by the FEM model. © 2005 Elsevier Ltd. All rights reserved.

Keywords: Buckling analysis | Composite tubes | FEM modelling | Lattice structures

Abstract: Computing how a camera samples a scene is a common task in engineering and computer graphics. Though this concept is rather intuitive (sampling density is proportional to the number of pixels which sample a point in the scene), it is necessary to define it in a formal and rigorous way. This paper describes a method to evaluate the sampling density distribution of a camera. A definition of camera sampling density is given, and a simple computation method is presented. The proposed approach is based on the evaluation of the area reduction coefficient associated to the function which maps a generic point of the scene to a point in the image plane. The approach can be applied either to the ideal (pinhole) camera model or to real camera models with distortion. Experimental tests on a specimen are also presented and discussed.

Keywords: Camera sampling density | Image acquisition | Reverse engineering

Abstract: The aim of this paper is to test a new infusion system prototype which is intended to recover the normal venous return and lost blood quantity in hypovolemic patients. For this purpose a bolus of fluid is injected at a high speed in the vena cava, accelerating and dragging the stagnating flow upstream the catheter. In order to improve the effects of the injection and to avoid damage to heart valves, the injection is synchronized with the heart cycle and should be controlled by the electrocardiogram track of the patient in such a way that the maximum flow rate injected occurs when the tricuspid valve is open. An in-vitro experimental apparatus - simulating the vena cava and its environment - has been built and used to measure velocity and pressure fields in the vena cava during the high velocity injection. In a first embodiment, the experimental apparatus consists of two reservoirs arranged, respectively, upstream and downstream a Starling Resistor, but at different heights. A constant flow rate inside the vena cava is maintained, thus simulating the diastolic phase of the heart. In a second embodiment of the experimental apparatus, two electrical valves, arranged downstream the Starling Resistor, generate an oscillating pressure wave along the vena cava, thus simulating the human central venous pressure. By varying the flow rate inside the vena cava and the opening rate of the valves, it has been possible to evaluate the dragging effect of the new injection system and the mechanical behavior of the vena cava during both continuous and pulsating infusion. Copyright © 2006 by ASME.

Abstract: Interoperability among CAD/CAM/CAE systems is a well known problem in product design and development. At present geometrical data exchange among different software packages is usually carried out through neutral file formats (IGES or STEP) or through proprietary formats. Data exchange processes are usually afflicted by several problems, such as: information loss, redundancy, one-way data exchange and static data exchange. These drawbacks do not permit a really geometric-centric design, and even if the model is transmitted without loss of information, the exchanged data do not incorporate details such as sketches, constraints and features, which represent the designer's intent. As a result, the model can hardly be modified, and the original intent of the designer may be misunderstood. During the last five years various solutions have been proposed to solve the above mentioned problems. The aim of this work is to investigate and discuss recent research trends in this topic. © 2006 IEEE.

Abstract: This work displays an integrated approach - based on both experimental and numerical analyses (CFD) - to scooter design. In the last decade the market of maxi scooters is continually increasing. Any constructor must develop new models, new design, new concepts. Due to this fast evolution, the possibility to cope aesthetics and functionality, at each step design process, may be a crucial aspects. The purpose of this work is the development of a method to take into account aerodynamic behaviour of a scooter depending on its external geometry. The analysis is mainly directed to investigate stability, safety and comfort aspects rather than aerodynamic fairness. The paper presents a methodology to integrate in the conceptual design process many aerodynamic aspects. First a parametric simplified model was created and investigated. This model is composed of few plane panels which outline a commercial scooter. The scope was the validation of the analysis method through an experimental campaign in a wind tunnel. Then, the methodology was applied to a real scooter. As a conclusion the paper points out the importance of embedding aerodynamic skills into the very initial conceptual design of a scooter, and the method proposed tries to answer to this need. Copyright © 2006 by ASME.

Abstract: A Design of Experiment (DoE) based methodology is proposed for use in optimising sailing mast structures. The method uses Finite Element Analysis to simulate mast behaviour when subject to typical design loadings. Due to the large number of possible combinations of design variables the use of a fractional factorial selection is used to determine their values for the limited number of simulations required. The analysis is based around a mast for use with the Mini Transat 6.50 class of yacht and provides a suitable challenge when compared to the typically conservative approach used in mast design. The mast structure is considered with respect to standing rigging working loads and spreaders geometry on the deformed shape of a pretensioned mast. Responses in terms of maximum value of mast camber, position of mast camber and generalized distance from optimal curves are provided, response surfaces over design variables domains are built and the most influential parameters are selected. Moreover, the present approach allowed us to provide combinations of parameters leading to optimal matching between a given target curve and the shape of a pretensioned mast. © 2006 Roval Institution of Naval Architects.

Abstract: The purpose of this research -based on CFD- is the evaluation of the aerodynamic performance of the geometry of a scooter. The analysis is mainly directed to investigate stability, safety and comfort aspects rather than aerodynamic fairness. A parametric simplified model was created and investigated. The model is composed of few plane panels which outline a commercial scooter. Pitching moment, velocity and turbulence fields are evaluated for different configurations and compared. The results are validated through a wind tunnel test campaign and are presented in the form of carpet plots which can drive the designer in his choices.

Keywords: Modelling | Motorcycles | Simulation

Abstract: This article describes a procedure based on reverse engineering and CFD analysis to numerically evaluate the aerodynamic performances of physical components. The case study is represented by the investigation of the aerodynamic behaviour of a racing car air-box. An existing air-box has been digitised using a proprietary system which enables the acquisition of a point cloud of the surface of the object. A procedure has been developed to obtain a CAD model suitable for CFD analysis from the raw 3D data. A computational fluid-dynamics analysis has been performed using the digital model. The procedure described in this paper can be easily extended to the reconstruction and the analysis of similar classes of shapes such as wings and profiles.

Keywords: Aerodynamic analysis | Geometric modelling | Reverse engineering

Abstract: The paper describes a method to optimise the thickness balance within a composite laminate with layers oriented according to a limited set of angles. The laminate must be symmetric, balanced and loaded in-plane. The optimisation process is particularly suited to be used in conjunction with a finite element program. It provides the designer with the optimal overall engineering characteristics and a list of all the possible orientation combinations ranked with respect to their safety factor. The optimisation method is based on the first order gradient optimum search method and operates iteratively in the engineering elastic characteristics field. The cost function implemented up to now is the structure stiffness but no conceptual limitations exist for different functions. Some simple applications are listed at the end of the paper in order to verify the capabilities of the method. © 1997 Elsevier Science Limited.

Abstract: The paper describes a method to optimize composite structures which is particularly useful in conjunction with the finite element technique. The main features of the method consist in a two level approach: first an overall optimization of an equivalent orthotropic laminate is performed, then an optimum lay-up sequence search is carried-on. The optimization techniques are based on the gradient method and the Lagrangian multipliers technique. In the present implementation some limitation must be fulfilled: the laminates must be balanced and symmetric. Up to now only stiffness related cost functions have been considered; some applications are displayed at the end of the paper.

Abstract: The paper describes a computer aided technique that allows automatic treatment of the photoelastic pattern obtained during photoelastic analysis of a loaded body. Both isoclinic and isochromatic information can be treated and the overall stress field can be obtained. The technique has been implemented into a computer program and tested.

Abstract: The aim of this paper is to investigate the influence of geometric parameters on the stress field generated in a laminate with a pin-loaded hole. As this work was developed in order to study the behavior of leaf-springs, particular emphasis was given to unidirectional laminates which are used in this kind of structure. However, some other laminates have also been studied-a ±45° angle-ply laminate and a quasi-isotropic laminate-in order better to understand the load transfer mechanism in a pin-loaded hole. The investigation was carried out with the aid of numerical methods and the results have been used in the Azzi-Tsai failure criterion to predict laminate strength. The geometric parameters considered are the distance from the center of the hole to the laminate edge and the laminate width. Some experimental tests were carried out to evaluate the effectiveness of the method and the results are given in the paper. © 1986.

Abstract: This paper relates an attempt to evaluate the interlaminar stresses that are generated within the interface between two layers near the free edge of angle-ply and cross-ply laminates. An approximate model based on simple polynomial formulations is proposed. The results obtained with this model are compared with numerical results available in the literature and with values obtained by an experimental program conducted on angle-ply glass/epoxy laminates.