Surface reconstruction based on CAD model driven priori templates.

To accurately quantify polyethylene wear in retrieved arthroplasty components, the original geometry of the component must be estimated accurately using a reference geometry such as a computer-aided design (CAD) model or a never-implanted insert. However, differences may exist between the CAD model and manufactured inserts resulting from manufacturing tolerances. We quantified the deviations between CAD models and newly manufactured inserts and determined how these deviations compared with using a never-implanted insert as a reference geometry. We obtained five cruciate-retaining (CR) and five posterior-stabilizing (PS) tibial inserts and their CAD models. The inserts were scanned and reconstructed using microcomputed tomography (micro-CT). Differences in volume and surface geometry were measured among (1) the individual inserts; (2) between the inserts and a CAD model; and (3) between the inserts and a reference geometry constructed from multiple scanned inserts averaged together. The micro-CT volumes were, on average, 0.4% smaller (34-178 mm(3)) than the CAD model volumes. The mean deviation between the CAD model and insert surface geometry was 25.7 μm smaller for CR and 36.8 μm smaller for PS. The mean deviation between the inserts and an averaged reference geometry was 1.4 μm larger for CR and 0.4 μm smaller for PS. Deviations exist between manufactured tibial inserts and CAD models that could cause errors in wear measurements. Scanned inserts may better represent the preimplantation geometry of worn inserts than CAD models, depending on the manufacturing variability between lots. The magnitude of the error in estimation of the preimplantation geometry of a retrieved component could add or subtract the equivalent of 1 year of wear.

The issue of improving quality, costs and delays indicators in design and manufacturing is more relevant than ever in the industry. After lean manufacturing, well known in production process, the lean engineering approach is being implemented today in the field of design, taking the name of lean product development. The management of knowledge and know-how (existing, new or to be acquired) is the heart of lean engineering. In our suggested methodology this is implemented through a new generation of tools called Knowledge Configuration Management (KCM) and Knowledge Extraction Assistant (KEA). KCM tools are lean engineering components that provide analytical approach to knowledge management and knowledge-based engineering. These tools require a highly integrated approach that involves, for example, predefined geometrical parametric 3D models, such as CAD templates. But this approach cannot be deployed in all engineering sites. We propose to complete this KCM approach introducing a semantic network approach, coupling with Feature Identity Card (FIC). FIC contains a set of metadata and information existing in the Product Data Management (PDM), connected with information extracted from 3D CAD (Computer Aided Design) models. It allows contextualizing information and ensures semantic connections, in order to manipulate the right parameters with mathematical algorithms. Those algorithms will search candidate relationships between design parameters extracted from CAD models. Our suggested approach aims at extracting knowledge in cases where design never came out of Knowledge Based Engineering (KBE) applications. In those situations, it seems important to complete classical knowledge management approach, and to find out the implicit knowledge embedded in 3D CAD models. This is achieved through a global approach, focusing on the product’s 3D definitions. We suggest introducing the latter approach by a suite of digital KEA tools (interfaced with KCM tools). Extracting knowledge from projects information stored in the Product Data Management does this. More precisely, the methodology is based on a commercial 3D similarity search tools for CAD models and on mathematical algorithms that search relationships between extracted design parameters. The goal is to submit new rules to the process and design experts. Implementing this methodology, a deeper knowledge of the product and its associated process can be acquired. This ensures a more productive and efficient design process.

Among machining conditions, tool paths exert a significant influence on the final machining outcome. In conventional practice, computer-aided design (CAD) models created from basic dimensions are typically used to generate tool paths within computer-aided manufacturing software. In recent years, 3D annotated models—CAD models enriched with product manufacturing information (PMI)—have become increasingly widespread. However, CAD models are often modified manually to achieve the desired machining results while accounting for machining errors. Consequently, operators consume time and effort in generating tool paths. Moreover, knowledge and experience in machining are required to modify CAD models according to PMI, such as unilateral tolerances. In this study, a method is proposed that leverages operator expertise to automatically modify CAD models and generate tool paths capable of satisfying specified size limits. The method employs a chain expression of dimensions, where the dimensional chain originates from the datum used as a reference for machining. Based on the dimensional chain, the objects requiring shifts are systematically identified. Subsequently, basic dimensions with unilateral tolerances are converted into target dimensions with bilateral tolerances to achieve the desired machining results. Case studies confirm that CAD models can be automatically modified when the target dimension is set to the median value of the size limits. Furthermore, machining experiments demonstrate that the proposed method effectively ensures compliance with the specified size limits.

We propose a Point-Voxel DeConvolution (PVDeConv) module for 3D data autoencoder. To demonstrate its efficiency we learn to synthesize high-resolution point clouds of 10k points that densely describe the underlying geometry of Computer Aided Design (CAD) models. Scanning artifacts, such as protrusions, missing parts, smoothed edges and holes, inevitably appear in real 3D scans of fabricated CAD objects. Learning the original CAD model construction from a 3D scan requires a ground truth to be available together with the corresponding 3D scan of an object. To solve the gap, we introduce a new dedicated dataset, the CC3D, containing 50k+ pairs of CAD models and their corresponding 3D meshes. This dataset is used to learn a convolutional autoencoder for point clouds sampled from the pairs of 3D scans - CAD models. The challenges of this new dataset are demonstrated in comparison with other generative point cloud sampling models trained on ShapeNet. The CC3D autoencoder is efficient with respect to memory consumption and training time as compared to stateof-the-art models for 3D data generation.

In this chapter, we address various issues related to data processing for rapid prototyping. In contrast to traditional material removal processes, rapid prototyping techniques build a part by incrementally adding materials layer by layer (Kruth 1991, Barequet and Kaplan 1998), so they are also called layer-based manufacturing or layered manufacturing. The processes are fully automatic and offer a number of competitive advantages over traditional manufacturing processes. Hence, they are particularly useful for rapid product development. The entire process for model prototyping is illustrated in Figure 5-1 and can be summarized as follows: CAD modeling: The first step in rapid prototyping is to prepare a computer-aided design (CAD) model of the object to be fabricated using layer based manufacturing processes. For most of the available RP technologies, a solid model with complete topological and geometric information is required. STL interfacing: STL is the de-facto standard for the rapid prototyping industry. It is a file format for approximately defining an object using triangular facets. When a CAD model is available, the entire part geometry is converted into STL format based on a tolerance for accuracy control. Part orientation: Before processing for prototyping, a RP engineer needs to figure out the specific orientation in which the prototype model will be produced. Part accuracy, the amount of supporting material required and ease of post-processing are important factors influencing part orientation determination. Support generation: Depending on the specific RP process to be used, one may need to further define support structures for supporting down-facing areas during part build-up. Support generation can be done on the basis of a STL model or the original CAD model. Model slicing and tool path generation: In contrast to material removal manufacturing technologies, rapid prototyping technologies refer to a class of layer-based material increase manufacturing processes. The digital model of the object and related support structures need to be sliced layer by layer. For each layer, a set of surface contours is obtained and a tool path is defined within the material area for model production in a layer-wise fashion. Model production on a RP machine: The produced tool path is sent to a RP machine for building up the prototype model, including support, layer by layer. Post-processing: Depending on the RP process involved, a post-processing step might be needed for post-curing in the case of stereolithography, for infiltration and furnace sintering in the case of SLS, and/or for removing the support structures and surface polishing in the cases of most other RP processes.

Nickel and nickel-alloy microparts sized on the order of 5–1000 microns have been imaged in three dimensions using a new microscopic technique, Digital Volumetric Imaging (DVI). The gears were fabricated using Sandia National Laboratories’ LIGA technology (lithography, molding, and electroplating). The images were taken on a microscope built by Resolution Sciences Corporation by slicing the gear into one-micron thin slices, photographing each slice, and then reconstructing the image with software. The images were matched to the original CAD (computer aided design) model, allowing LIGA designers, for the first time, to see visually how much deviation from the design is induced by the manufacturing process. Calibration was done by imaging brass ball bearings and matching them to the CAD model of a sphere. A major advantage of DVI over scanning techniques is that internal defects can be imaged to very high resolution. In order to perform the metrology operations on the microcomponents, high-speed and high-precision algorithms are developed for coordinate metrology. The algorithms are based on a least-squares approach to data registration the {X,Y,Z} point clouds generated from the component surface onto a target geometry defined in a CAD model. Both primitive geometric element analyses as well as an overall comparison of the part geometry are discussed. Initial results of the micromeasurements are presented in the paper.

This work deals with development of algorithm for physical replication of patient specific human bone and construction of corresponding implants/inserts RP models by using Reverse Engineering approach from non-invasive medical images for surgical purpose. In medical field, the volumetric data i.e. voxel and triangular facet based models are primarily used for bio-modelling and visualization, which requires huge memory space. On the other side, recent advances in Computer Aided Design (CAD) technology provides additional facilities/functions for design, prototyping and manufacturing of any object having freeform surfaces based on boundary representation techniques. This work presents a process to physical replication of 3D rapid prototyping (RP) physical models of human bone from various CAD modeling techniques developed by using 3D point cloud data which is obtained from non-invasive CT/MRI scans in DICOM 3.0 format. This point cloud data is used for construction of 3D CAD model by fitting B-spline curves through these points and then fitting surface between these curve networks by using swept blend techniques. This process also can be achieved by generating the triangular mesh directly from 3D point cloud data without developing any surface model using any commercial CAD software. The generated STL file from 3D point cloud data is used as a basic input for RP process. The Delaunay tetrahedralization approach is used to process the 3D point cloud data to obtain STL file. CT scan data of Metacarpus (human bone) is used as the case study for the generation of the 3D RP model. A 3D physical model of the human bone is generated on rapid prototyping machine and its virtual reality model is presented for visualization. The generated CAD model by different techniques is compared for the accuracy and reliability. The results of this research work are assessed for clinical reliability in replication of human bone in medical field.

Surface reconstruction by parallel and unified particle-based resampling from point clouds

4D CAD model updating using image processing-based construction progress monitoring

This in vitro study evaluated the dimensional accuracy of three 3D printers and one milling machine with their respective polymeric materials using a simplified geometrical model. A simplified computer-aided design (CAD) model was created. The test samples were fabricated with three 3D printers: a dental desktop stereolithography (SLA) printer, an industrial SLA printer, and an industrial fused deposition modeling (FDM) printer, as well as a 5-axis milling machine. One polymer material was used per industrial printer and milling machine while two materials were used with the dental printer for a total of five study groups. Test specimens were then digitized using a laboratory scanner. The virtual outer caliper method was used to measure the linear dimensions of the digitized 3D printed and milled specimens in x-, y-, and z-axes, and compare them to the known values of the CAD model. Data were analyzed with Kruskal-Wallis one-way ANOVA on Ranks followed by the Tukey's test. Milled specimens were not significantly different from the CAD model in any dimension (p > 0.05). All 3D printed specimens were significantly different from the CAD model in all dimensions (p = 0.01), except the dental SLA 3D printer with one of the polymers tested (Bis-GMA) which was not significantly different in two (x and z) dimensions (p = 0.4 and p = 0.12). The milling technology tested provided greater dimensional accuracy than the selected 3D printing. Printer, printing technology, and material selection affected the accuracy of the printed model.

Digital manufacturing (DM) is actively adopted to the production lifecycles of a variety of critical industries, and this rapid growth has resulted in exponential increase of 3D computer-aided design (CAD) models. Unfortunately, counterfeiting of intellectual property becomes a prominent threat as many 3D designs are accessible online, combined with the proliferation of cheap consumer 3D printers that enable malicious actors to produce non-authentic parts. State-of-the-art techniques to secure manufacturing processes mostly rely on watermarking, which embeds hidden information inside CAD models to prove ownership and authenticity. Nevertheless, such techniques tamper with the model itself, while existing attacks allow removing such watermarks altogether. To address these shortcomings, we integrate signal processing and cryptographic techniques and describe a tailored solution for CAD model ownership and supply chain management. Our approach generates unique identifiers for 3D designs using frequency-domain transforms and employs non-fungible tokens (NFTs) that persist on public distributed ledgers. Our NFTs are implemented on the Ethereum blockchain using smart contracts and their functionality is twofold: (a) authenticate the owner of a CAD model, and (b) enable ownership transfer. To validate our technique, we deployed our smart contract on Ethereum's proof-of-work Ropsten network and demonstrated the applicability of our methodology.

An intelligent approach for dimensioning completeness inspection in 3D based on transient geometric elements

Herein, we propose a novel indoor structure extraction (ISE) method that can reconstruct an indoor planar structure with a feature structure map (FSM) and enable indoor robot navigation using a navigation structure map (NSM). To construct the FSM, we first propose a two-staged region growing algorithm to segment the planar feature and to obtain the original planar point cloud. Subsequently, we simplify the planar feature using quadtree segmentation based on cluster fusion. Finally, we perform simple triangulation in the interior and vertex-assignment triangulation in the boundary to accomplish feature reconstruction for the planar structure. The FSM is organized in the form of a mesh model. To construct the NSM, we first propose a novel ground extraction method based on indoor structure analysis under the Manhattan world assumption. It can accurately capture the ground plane in an indoor scene. Subsequently, we establish a passable area map (PAM) within different heights. Finally, a novel-form NSM is established using the original planar point cloud and the PAM. Experiments are performed using three public datasets and one self-collected dataset. The proposed plane segmentation approach is evaluated on two simulation datasets and achieves a recall of approximately 99%, which is 5% higher than that of the traditional plane segmentation method. Furthermore, the triangulation performance of our method compared with the traditional greedy projection triangulation show that our method performs better in terms of feature representation. The experimental results reveal that our ISE method is robust and effective for extracting indoor structures.

At present, the learning-based down-sampling method has achieved good result by using the loss of subsequent tasks to optimize the process of sampling points selection. However, for the current methods there still has two problems, on the one hand, the sampled points may not be in the original point cloud, and the sampled generated points of the sample need to be matched with the original point cloud (re-matching problem), on the other hand, the number of points obtained after matching may be insufficient for the sampling demand, which need to be supplemented by Farthest Point Sampling, nearest neighbor point, or other methods (re-supplement problem). In order to solve the above problems of re-matching and re-supplement in the learning-based sampling method, we propose a new sampling network which based on high score selection strategy. In our method, LSTM (Long Short Term Memory) is used as the feature extraction method to obtain the relational weight which relative to the subsequent tasks. The attention mechanism is used to select the high score points. The experimental results show that the performances of our network (We dubbed it as LA-Net) are better than other sampling methods and achieve the state-of-the-art in the unified benchmark. Detailed ablation studies have also been carried out, which provably affirm the advantage of each component we proposed. Finally, the methods in this paper are applied to a real backbone network to optimize its performance, and real robot grasping experiments are designed to realize automatic robot grasping of multi-target objects in real scenes.

Point cloud is the popular researches in the industry, and the research on partial point cloud is favored by many researchers. In the acquisition process of original 3D point cloud data, there are partial or sparse problems due to occlusion, lighting and other reasons, which will lead to deviation in downstream tasks. We propose a network MDPCN using multiple decoders, which can make the partial point cloud complete. The network uses deep learning to get multiple features of the partial point cloud, and uses multiple identical decoders to decode. Each decoder uses the self-attention module and the Folding-Net to generate smooth point clouds from the features, and finally integrates the point clouds generated by each decoder to obtain dense point cloud. Compared with other mainstream point cloud completion network methods and ablation experiments, it can be proved that this network model can generate more accurate point cloud and effectively complete the 3D reconstruction task.