Publications

This research work deals with the geometric modelling of 5-axis machine tool based on a standard-ised parameterisation of geometric errors with the aim to decrease the volumetric error in the workspace. The identification of the model's parameters is based on the development of a new standard thermo-invariant material namely the Multi-Feature Bar. Thanks to its calibration and a European intercom-parison, it now provides a direct metrological traceability to the SI metre for dimensional measurement on machine tool in a hostile environment. The identification of three intrinsic parameters of this standard , coupled with a measurement procedure ensures a complete and traceable identification of motion errors of linear axes. An identification procedure of location and orientation errors of axes is proposed by probing a datum sphere in the workspace and minimising the time drift of the structural loop and the effects of the previously identified motion errors. Finally, the developed model partially identified, allows the characterisation of 95% of the measured volumetric error. Therefore, the mean volumetric error not characterised by the model only amounts to 8 µm.

Even if 3D acquisition systems are nowadays more and more e cient, the resulting point clouds nevertheless contain quality defects that must be taken into account beforehand, in order to better anticipate and control their e ects. Assessing the quality of 3D acquisitions has therefore become a major issue for scan planning. This paper presents several quality metrics that are then studied to identify those that could be used to optimize the acquisition positions to perform an automatic scan. From the experiments, it appears that, when considering multiple acquisition positions, the coverage ratio and score indicator have signi cant changes and can be used to evaluate the quality of the measurements. Di erently, other indicators such as e cacy ratio, registration

Accurately transferring the real world to the virtual one through reverse engineering is of utmost importance in Industry 4.0 applications. Indeed, acquiring good quality 3D representations of existing physical objects or systems has become mainstream to maintain the coherence between a real object and its digital twin. Compared with traditional contact measurement, contact-less scanning is undoubtedly a fast and direct acquisition technology. However, for a given acquisition, finding the right scanning configuration remains a challenging question whose resolution has attracted researchers in recent years. Using heuristics and visibility criteria, some approaches try to automatically plan the positions and path to be followed by a robot when scanning an object being manufactured [1]. Similarly, Joe Eastwood et al. use a genetic algorithm and a convolutional neural network to optimize the locations of the cameras with the purpose that maximize surface coverage and measurement quality [2]. However, all those techniques base their reasoning on theoretical models whose real behavior may diverge as compared to real measuring. Thus, being able to take decisions based on the results obtained from real acquisitions is crucial to minimize the deviations between what was planned and what has been obtained by the end. To do so, ad-hoc metrics need to be used to accurately characterize the quality of point clouds that are then used in the next engineering steps (e.g. reconstruction, control, simulation). The methods for evaluating point cloud (PC) quality can be divided into two types, i.e. subjective and objective. The former mainly evaluates the point cloud from a perceived visual quality for immersive representation of 3D contents [3][4], whereas the latter is more quantitatively based on values. For quantitative metrics for evaluating the quality of PC, some researchers only considered the properties of the PCs, assessing the qualities of the PC from four aspects [5]: noise, density, completeness, and accuracy of the point cloud data. Based on these achievements, some scholars further proposed an indicator for surface accessibility, to characterize how a region on the surface of the workpiece can be reached or not by the scanner. Besides, the coverage rate was proposed to reveal how much the area is scanned. Additionally, the normal angle error was figured out in [4]. However, all those metrics can behave differently depending on the adopted technology: laser scanner, photogrammetry, or structured- light measuring for instance. Catalucci et al. [6] compared the photogrammetry and structedlight measurements on additively manufactured parts and proposed quality indicators of PC that include measurement performance indicators and statistical indicators on the whole part measurement. However, their work focused on whole scans of the part that consist of many point clouds acquired from different scan positions and configurations. Although many criteria have been proposed, it remains to be investigated which are the most accurate and obvious metrics to evaluate the quality of the point cloud during a structured light-based scan

The fourth industrial revolution is forcing companies to rethink their status quo – creating a need to assess their digital maturity as a basis for improvements. As a result, there is a variety of maturity models available in the literature. This paper introduces a novel comparison framework designed to compare different digital maturity assessment models. Our framework has several steps: reverse engineering of criteria from existing models, criteria matching analysis, as well as computation of the coverage and spread ratios. These two metrics characterize respectively the similarity of two maturity models, and the spread between them. We tested the proposed approach with two well-known maturity self-assessment approaches, namely the IMPULS and PwC methods. From our analysis, we were able to derive several insights that will help to develop a new maturity model specifically dedicated to support SMEs in the aerospace industry and manufacturing sector.

Natural fiber reinforced polymers (NFRPs) are environmentally friendly and are receiving growing attention in the industry. However, the multi-scale structure of natural fibers and the random distribution of the fibers in the matrix material severely impede the machinability of NFRPs, and real-time monitoring is essential for quality assurance. This paper reports a synchronous in situ imaging and acoustic emission (AE) analysis of the NFRP machining process to connect the temporal features of AE to the underlying dynamics and process instability, all happen within milliseconds during the NFRP cutting. This approach allows directly observing the surface modification and chip formation from a high-speed camera (HSC) during NFRP cutting processes. The analysis of the HSC images suggests that the complex fiber structure and the random distribution introduce an unsteady, almost a freeze-and-release type motion pattern of the cutting tool with varying depths of cut at the machining interface. More pertinently, a prominent burst pattern of AE from time domain was found to emanate due to the sudden penetration of the tool into the surface of the NFRP workpiece (increasing the depth of cut), as well as a release motion of the tool from its momentary freeze position. These findings open the possibility of tracking AE signals to assess the effective specific energy and surface quality that are affected by these unsteady motion patterns.

AbstractThis paper addresses the way a simulated annealing-based fitting strategy can be enhanced by leveraging a sensitivity analysis able to characterize the impact of the variations in the parameters of a CAD model on the evolution of the deviation between the CAD model itself and the point cloud of the digitized part to be fitted. The principles underpinning the adopted fitting algorithm are briefly recalled. The applied sensitivity analysis is described together with the comparison of the resulting sensitivity evolution curves with the changes in the CAD model parameters imposed by the simulated annealing algorithm. This analysis suggests several possible improvements that are discussed. The overall approach is illustrated on the fitting of single mechanical parts but it can be directly extended to the fitting of parts’ assemblies. It is particularly interesting in the context of the Industry 4.0 to update digital twins of physical products and systems.

Enlarging 3D model databases by shape synthesis is a large field of research. Indeed, the use of machine learning techniques requires a huge amount of labeled CAD models, and it is therefore crucial to rely on large and varied databases. Most of existing works in shape synthesis focus on everyday life objects generation [1]. However, these methods often do not work on assemblies composed of several CAD models, and it is the aim of this paper to develop a new shape synthesis method to enlarge existing CAD assembly databases. Today, there exist lots of free databases of non-labeled CAD models (e.g. GrabCAD, 3D Warehouse, Turbosquid) often available as STEP or IGES files. Unfortunately, very few of these databases are labeled. Other databases like PartNet [2] and ShapeNet [3] are currently labeled by crowdsourcing, but they do not contain complex mechanical assemblies. Current works in shape synthesis often use auto-encoders to generate new coherent CAD assemblies [4]. Moreover, there are probabilistic models to create diversity in large 3D Database [5] or to classify 3D assemblies [6]. Those techniques often use linkage graphs [7] to classify and generate new coherent assemblies. Furthermore, information within the linkage graphs differs according to the method, and those graphs are not suitable for complex CAD assemblies like hydraulic pumps. But the main issue of all methods is still that those databases have to be labelled. The method explained in this paper consists in creating new labeled CAD assemblies from existing ones by linkage graph overlay. Here, the STEP file format has been adopted in order to be the most reproductible and to be adaptable. The linkage graphs are automatically created thanks to the identification of the linkages between the components. Indeed, linkages are not included in the STEP files and they need to be computed. Theses linkage graphs are then analyzed and components with similar linkages are detected. Finally, once the similarities detected, the corresponding components can be exchanged to created new assemblies for which the labels can be directly inherited from the source assemblies. The contribution is threefold: (i) a method to create linkage graphs from existing non-labelled CAD assemblies; (ii) a method to recognize basic components using linkage graphs; (iii) a smart overlay method to replace some components while keeping the coherence between all the components of the assembly. The algorithm has been implemented in Python on FreeCAD and it has been tested on several test cases. Figure 1 shows the overview of the method, from the graph synthesis to the components overlay, finishing with the replacement of the components. The results are presented and discussed, and a conclusion ends this extended abstract while discussing the next steps.

Abstract. The purpose of this study is to enhance point cloud semantic segmentation by using point clouds from multiple distinct technologies on the same capture location and to determine whether employing various technologies throughout the acquisition process yields better performance during classification. The different point clouds were captured in the same geographical location and have previously been aligned and classified by professionals of the field. Three locations have been scanned with airborne lidar, terrestrial lidar and photogrammetry using UAV or helicopter. The use of various sources of capture on the same location opens the door to creating new features, such as the proportion of each source involved in the semantic segmentation of point clouds. This plurality of sources also enables us to spread various features, such as RGB colors, that have been propagated to other sources via the neighborhood. The initial results lean towards capture using different technologies as the overall accuracy increase by two to four points and the mean Matthews correlation coefficient increase by four to seven points. The main drawbacks are the cost of some technologies, as well as the processing time, which is greater than with a single technology.

Les travaux présentés ici visent à modéliser le comportement dynamique des machines de forgeage, afin de considérer à terme, l’influence du comportement des machines sur le procédé. Pour se faire, l’étude se limite au cas des machines pilotées en énergie, et l’ensemble {machine + outillages} est considéré. De plus, les modèles développés s’attachent à modéliser le comportement de l’ensemble uniquement pendant la phase de frappe. L’objectif est de rendre les simulations plus prédictives en intégrant l’efficacité de la machine. Ces avancées auront pour conséquence une meilleure connaissance du chemin thermomécanique subi par la pièce dès la simulation. Ce qui contribuera à l’optimisation des gammes de forgeage pour la maîtrise de la microstructure des matériaux et de la qualité des pièces forgées.

Machining of biocomposites using traditional techniques has shown some limitations due to the multiscale complex cellulosic structure of natural fibrous reinforcement. This paper aims to demonstrate the feasibility of the cryogenic nitrogen jet, which is a novel cutting process that combines sustainable resources and cryogenic temperatures from −175 °C to 150 °C, as a machining process for biocomposites made of unidirectional flax fibers and polylactic-acid polymer (PLA). A high-velocity stream of liquid nitrogen with and without abrasives (400–700 m/s) is hence directed onto the workpiece surface. For the abrasive jet, both conventional garnet abrasives and bio-based abrasives made from walnut shells were used for the nitrogen jet process. The kerf depth, which is representative of the erosion wear mechanisms, was calculated to assess the erosion rate of the biocomposite structure after the nitrogen jet cutting operation. Then, scanning electron microscope observations are performed to characterize the induced damages on the machined biocomposite surfaces. Results show that the pressure and the traverse speed of the nitrogen jet are the relevant process parameters that control the mechanical and thermal stresses viewed by the biocomposite during the machining operation. Each impinging abrasive particle removes a small amount of biocomposite material during the jet cutting process, which depends on the brittleness of the anisotropic work structure and the abrasiveness of the incorporated particles into the jet stream.