Pose Estimation for Objects with Rotational Symmetry

Inaccurate results are usually produced due to lack of precise dynamic parameters when analyzing the running stability of railway vehicles. In this paper, a new solution is proposed to conduct the running stability analysis of railway vehicles supported by their three-dimensional (3D) CAD models, which are realized based on the intrinsic mapping relationship between the dynamic model and the 3D CAD model of railway vehicles. First, a dynamic model library of railway vehicles is established, and for each model type the software package for running stability analysis is programmed with Matlab. Then according to the requested parameters of each model, the corresponding dynamic parameter extraction flow based on 3D CAD model is designed by the secondary development of CAD software. The user chooses CAD objects and sets part of dynamic parameters using a wizard-style guiding dialogs; the background program will automatically and accurately extract the physical and geometric parameters from the CAD model, and then call the Matlab software package and transfer the parameter values, with analysis results presented in CAD environment. The case study shows that, this method can not only ensure the precision of dynamic parameters, but also improve the dynamic modeling efficiency for running stability analysis of railway vehicles.

FuS-GCN: Efficient B-rep based graph convolutional networks for 3D-CAD model classification and retrieval

Conception et fabrication de moulages assistées par ordinateur

Feature-based NC machining, which requires high quality of 3D CAD model, is widely used in machining aircraft structural part. However, there has been little research on how to automatically detect the CAD model errors. As a result, the user has to manually check the errors with great effort before NC programming. This paper proposes an automatic CAD model errors detection approach for aircraft structural part. First, the base faces are identified based on the reference directions corresponding to machining coordinate systems. Then, the CAD models are partitioned into multiple local regions based on the base faces. Finally, the CAD model error types are evaluated based on the heuristic rules. A prototype system based on CATIA has been developed to verify the effectiveness of the proposed approach. Highlights The 3D CAD model of an aircraft structural part is partitioned into multiple local regions in terms of the base faces. The fillet association level graph is proposed to represent the local region for exploring the CAD model errors. An automatic CAD model error detection method based on heuristics rule is proposed.

Object pose estimation is an important problem in robotics because it supports scene understanding and enables subsequent grasping and manipulation. Many methods, including modern deep learning approaches, exploit known object models, however, in industry these are difficult and expensive to obtain. 3D CAD models, on the other hand, are often readily available. Consequently, training a deep architecture for pose estimation exclusively from CAD models leads to a considerable decrease of the data creation effort. While this has been shown to work well for feature-and template-based approaches, real-world data is still required for pose estimation in clutter using deep learning. We use synthetically created depth data with domain-relevant background randomized noise heuristics to train an end-to-end, multi-task network, for pose estimation. We simultaneously detect, classify and estimate the poses of texture-less objects in cluttered real-world depth images of an arbitrary amount of objects. We present the results of our experiments with the LineMOD and the Occlusion dataset.

Enhancing CNN-based network intrusion detection through hyperparameter optimization

The state-of-art 6D object pose detection methods use convolutional neural networks to estimate objects' 6D poses from RGB images. However, they require huge numbers of images with explicit 3D annotations such as 6D poses, 3D bounding boxes and 3D keypoints, either obtained by manual labeling or inferred from synthetic images generated by 3D CAD models. Manual labeling for a large number of images is a laborious task, and we usually do not have the corresponding 3D CAD models of objects in real environment. In this paper, we develop a keypoint-based 6D object pose detection method (and its deep network) called Object Keypoint based POSe Estimation (OK-POSE). OK-POSE employs relative transformation between viewpoints for training. Specifically, we use pairs of images with object annotation and relative transformation information between their viewpoints to automatically discover objects' 3D keypoints which are geometrically and visually consistent. Then, the 6D object pose can be estimated using a keypoint-based geometric reasoning method with a reference viewpoint. The relative transformation information can be easily obtained from any cheap binocular cameras or most smartphone devices, thus greatly lowering the labeling cost. Experiments have demonstrated that OK-POSE achieves acceptable performance compared to methods relying on the object's 3D CAD model or a great deal of 3D labeling. These results show that our method can be used as a suitable alternative when there are no 3D CAD models or a large number of 3D annotations.

Air gears must meet high performance criteria for accuracy; therefore, quality control is necessary for their production. CMM Software determines the capabilities and scope of the measuring machines. Measuring machines equipped with specialized software for the measurement of the gears are commercially available. An example might be a company offering Wenzel TGEAR XY software. This specialized software for the measurement of gears allows fully automatic measurement of the contour deviation (involute), line tooth pitch, and radial run out, tooth flank topography. If you do not have a special software to measure the gear, you require a measurement process based on 3D CAD model of the gear. It is necessary in this case, the preparation of a strategy based on measurements of 3D CAD model and load points or paths with the CAD model. The results of the measurement refer to the CAD model treating it as the nominal model. Item will present the methodology of measurement of cylindrical gears using a standard CMM software. Cylindrical gears can be measured through non-contacting or contacting methods – measuring point or scanning. The choice of measurement method should be carried out taking into account the accuracy of gear and taking into account the capabilities of the system and measurement needs. An important advantage of the measurement using the gears multipurpose measuring software is that it is possible to measure any geometry of the wheels in relation to the CAD model. Using this measurement strategy avoids the need to purchase additional software, the measuring machine and its setup. You should only choose a method of analysis that will be carried out directly in the environment-measuring machine, or will be made in the CAD system based on the geometry acquired during the measurement with respect to the virtual model. In the second case, export is needed in the appropriate format measurement data to CAD software.

This paper proposes a new methodology for robotic offline programming (OLP) addressing the issue of automatic program generation directly from 3D CAD models and verification through online 3D reconstruction. Limitations of current OLP include manufacturing tolerances between CAD and workpieces and inaccuracies in workpiece placement and modelled work cell. These issues are addressed and demonstrated through surface scanning, registration, and global and local error estimation. The method allows the robot to adjust the welding path designed from the CAD model to the actual workpiece. Alternatively, for non-repetitive tasks and where a CAD model is not available, it is possible to interactively define the path online over the scanned surface.KeywordsRoot Mean Square ErrorControl PointRobot TrajectoryWelding PathVisibility ConstraintThese keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

This abstract encapsulates a transformative summer internship at Fermilab, focusing on the Mu2e Experiment within the Mechanical Integration and 3D CAD Modeling group. The internship was characterized by a two-phase approach, with the initial month dedicated to the design of an aluminum platform tailored for DS Trench Maintenance. This phase encompassed a comprehensive exploration of structural integrity, material selection, and precision engineering principles, demonstrating an adeptness in CAD modeling and a keen eye for detail. The subsequent month saw a seamless transition into a dynamic phase of the project, wherein the intern played a pivotal role in effecting numerous adjustments across diverse sections of the 3D experiment structure. This phase demanded a nuanced understanding of the intricacies of the Mu2e Experiment and a high level of proficiency in CAD modeling.

Loops are critical elements in boundary representation (B-rep) models because they link all edges corresponding to a face. Loops can be used in feature recognition for identifying depressions or protrusions. In real 3D CAD models, however, features typically lie across multiple faces, which is beyond the data structure of current B-rep models. This study presents a virtual loop concept to account for all loop types used in CAD models, and develops algorithms for recognizing them. In accordance with the complexity of the recognition algorithm, this study defines three types of loop: single, virtual, and multivirtual. A single loop is the current loop recorded in the B-rep model. A virtual loop lies across faces that are at least G1 continuous. Finally, a multivirtual loop lies across faces that are either G0 or G1 continuous. The proposed loop structure provides a more complete data structure for recognizing various types of features in feature-recognition modules. Several realistic CAD models are presented to confirm the feasibility of the proposed loop-recognition and feature-recognition methods.

Bin-picking is a practical and challenging robotic manipulation task, where accurate 6D pose estimation plays a pivotal role. The workpieces in bin-picking are typically texture-less and randomly stacked in a bin, which poses a significant challenge to 6D pose estimation. Existing solutions are typically learning-based methods, which require object-specific training. Their efficiency of practical deployment for novel workpieces is highly limited by data collection and model retraining. Zero-shot 6D pose estimation is a potential approach to address the issue of deployment efficiency. Nevertheless, existing zero-shot 6D pose estimation methods are designed to leverage feature matching to establish point-to-point correspondences for pose estimation, which is less effective for workpieces with textureless appearances and ambiguous local regions. In this paper, we propose ZeroBP, a zero-shot pose estimation frame-work designed specifically for the bin-picking task. ZeroBP learns Position-Aware Correspondence (PAC) between the scene instance and its CAD model, leveraging both local features and global positions to resolve the mismatch issue caused by ambiguous regions with similar shapes and appearances. Extensive experiments on the ROBI dataset demonstrate that ZeroBP outperforms state-of-the-art zero-shot pose estimation methods, achieving an improvement of 9.1 % in average recall of correct poses.

Generally for any new design initially a concept layout in 2D CAD is generated. But sometimes, if the geometry of the product becomes complicated, then 3D CAD model is preferred, because 3D CAD model can give us a complete all side view at a time like a real product, but in a virtual world. The present case study will show the utilisation of 3D CAD at the concept design stage of a complicated shaped product for a new system. This will also give an idea about cost and time comparison. This will also show how 3D CAD model can also helps to analyse the product. Thus, this paper will describe about the importance of 3D CAD tools for product development.

This study proposes a method for automating the determination of assembly order by automating the derivation of the necessary connection relationships between the parts. The proposed method minimizes the information required for the initial conditions and automatically determines the feasible assembly orders. As a general rule, based on the assumption that the assembly order for a product is the reverse of the disassembly order, once the disassembly order is derived based on the 3D CAD model and the connection relationships between the parts, the assembly order can be determined. Until now, however, the relationships between the parts are decided manually by the attendant engineers, thus, hindering the full automation of the determination of the assembly order. To achieve full automation realistically, the connection relationships between the parts should be derived automatically from the 3D CAD model, for which this study proposes an efficient method. The components were extracted from the 3D CAD model, and the bolts were identified. The connection relationships between the parts were derived from the interference conditions determined while moving each part minutely. An association chart diagram was created from the obtained connection relationships, from which multiple assembly order candidates could be derived.

In order to improve the classification effect of the 3D CAD model, this paper combines the knowledge recognition algorithm of convolutional neural network to construct the 3D CAD model classification model. Moreover, this paper analyzes the application of the convolutional neural network algorithm in the classification of 3D CAD models, improves the knowledge recognition algorithm, and builds the basic structure of the convolutional neural network. In addition, this paper builds an intelligent recognition model with the support of the improved algorithm, designs experiments to verify the effect of this paper, and draws a visual display statistical graph based on the statistical test results. Through experimental research, it can be known that the 3D CAD classification model based on the knowledge recognition algorithm of convolutional neural network proposed in this paper can effectively improve the classification effect of the CAD model.