Feature reasoning for automatic robotic assembly and machining in polyhedral representation†

The automatic assembly of photo products has been developed by Nikon Corp. over more than ten years. Automatic assembly methods are historically classified into three categories: dedicated machines, semi-flexible automatic assembly systems, and universal automatic assembly systems. This paper surveys these types of automatic assembly systems. First, automatic assembly systems for mechanical components of photo products are considered, and their background and design policies discussed. Next, the performances of these assembly systems and the results achieved in practise are described. Finally, the future trends of photo product automation are outlined.

Error compensation of industrial robot based on deep belief network and error similarity

Future robotic assembly systems will allow product designers to upload their assembled product models remotely such that they can be assembled autonomously. In this work we present the architecture for a Click and Assemble Robotic Assembly (CARA) system. This architecture takes an assembly file uploaded by the user through a web interface as the only input. It then performs all the necessary planning before executing the assembly. To the authors’ knowledge, this is the first time that all of the required components from previous advances in robotic assembly have been brought together, with all interconnecting challenges solved, into a complete working system for physical, real world assembly tasks. To demonstrate the implemented architecture capabilities, a real industrial product with tight tolerances was assembled from its CAD file only, illustrating the end to end robotic assembly premise in a real world setting.

Retrieving images from a large image collection has been an active area of research. Most of the existing works have focused on content representation. In this paper, we address the issue of identifying relevant images quickly. This is important in order to meet the users' performance requirements. We propose a framework for fast image retrieval based on object shapes extracted from objects within images. The framework builds a hierarchy of approximations on object shapes such that shape representation at a higher level is a coarser representation of a shape at the lower level. In other words, multiple shapes at a lower level can be mapped into a single shape at a higher level. In this way, the hierarchy serves to partition the database at various granularities. Given a query shape, by searching only the relevant paths in the hierarchy, a large portion of the database can thus be pruned away. We propose the angle mapping (AM) method to transform a shape from one level to another (higher) level. AM essentially replaces some edges of a shape by a smaller number of edges based on the angles between the edges, thus reducing the complexity of the original shape. Based on the framework, we also propose two hierarchical structures to facilitate speedy retrieval. The first, called Hierarchical Partitioning on Shape Representation (HPSR), uses the shape representation as the indexing key. The second, called Hierarchical Partitioning on Angle Vector (HPAV), captures the angle information from the shape representation. We conducted an extensive study on both methods to see their quality and efficiency. Our experiments on sets of images, each of which has objects around from 1 to 30, showed that the framework can provide speedy image retrieval without sacrificing on the quality. Both proposed schemes can improve the efficiency by as much as hundreds of times to sequential scanning. The improvement grows as image database size, objects per image or object dimension increase.

Understanding 3D object shapes necessitates shape representation by object parts abstracted from results of instance and semantic segmentation. Promising shape representations enable computers to interpret a shape with meaningful parts and identify their repeatability. However, supervised shape representations depend on costly annotation efforts, while current unsupervised methods work under strong semantic priors and involve multi-stage training, thereby limiting their generalization and deployment in shape reasoning and understanding. Driven by the tendency of high-dimensional semantically similar features to lie in or near low-dimensional subspaces, we introduce a one-stage, fully unsupervised framework towards semantic-aware shape representation. This framework produces joint instance segmentation, semantic segmentation, and shape abstraction through sparse representation and feature alignment of object parts in a high-dimensional space. For sparse representation, we devise a sparse latent membership pursuit method that models each object part feature as a sparse convex combination of point features at either the semantic or instance level, promoting part features in the same subspace to exhibit similar semantics. For feature alignment, we customize an attention-based strategy in the feature space to align instance- and semantic-level object part features and reconstruct the input shape using both of them, ensuring geometric reusability and semantic consistency of object parts. To firm up semantic disambiguation, we construct cascade unfrozen learning on geometric parameters of object parts. Experiments conducted on benchmark datasets confirm that our approach results in instance- and semantic-level joint segmentation and shape abstraction with repeatable primitives, providing coherent semantic interpretations of 3D object shapes across categories in a one-stage, fully unsupervised manner, without relying on annotations or heuristic semantic priors.

The engine is one of the key core components in aircraft, in order to enhance automation and efficiency in the process of installation and disassembly for current small engine, using automatic control technology, a set of automatic disassembly and assembly system for aircraft engines is developed to replace the traditional mechanical installation vehicle. Corresponding for small engine to study the automatic disassembly system, the system uses laser sensors to locate the engine disassembly and assembly. Servo motors are used to drive the walking mechanism, lifting mechanism, and fine-tuning mechanism to realize multi-degree-of-freedom adjustment during engine disassembly and assembly. And the digital video is used to monitor the gap collision. Finally, the paper realizes the automatic high-precision disassembly and assembly of small engines. Through verification, the efficiency of the disassembly and assembly system has increased by more than 70%, the work efficiency has been greatly improved.

In this paper, type II robotic mixed-model two-sided assembly line balancing (RMTALB-II) problem is considered. In recent years, robots have been widely used in assembly systems as called robotic assembly lines where a set of tasks have to be assigned to stations and each station needs to select one of the different robots to process the assigned tasks. Two-sided assembly lines are especially used to produce large-sized high-volume products, such as automobiles, trucks, and buses. In this type of production line, both left side and right side of the line are used in parallel. However, little attention has been paid to solve RMTALB problems. Moreover, according to our best knowledge, there is no published work in the literature on RMTALB. This paper presents a new mixed-integer programming model for RMTALB-II to minimize the cycle time for a given number of mated stations. Since RMTALB problems are in NP-hard class of combinatorial optimization problems, a simulated annealing (SA) algorithm as metaheuristic method is proposed to solve the problem. Some problems are solved by SA and performance of the SA is evaluated by comparing their results with the optimal solution. The computational results show that SA is more efficient.

In image retrieval, relevance feedback uses information, obtained interactively from the user, to understand the user's perceptions of a query image and to improve retrieval accuracy. We propose simultaneous relevant feature selection and classification using the samples provided by the user to improve retrieval accuracy. The classifier is defined by a separating hyperplane, while the sparse weight vector characterizing the hyperplane defines a small set of relevant features. This set of relevant features is used for classification and can be used for analysis at a later stage. Mutually exclusive sets of images are shown to the user at each iteration to obtain maximum information from the user. Experimental results show that our algorithm performs better than the feature relevance weighting and feature selection schemes and comparably with the classification scheme using SVMs, in terms of retrieval accuracy, and it has the advantage of being faster than the classification scheme using SVMs.

In this paper, we present an efficient assembly sequence generation method for robot programming, which can reduce the complexity of assembly sequence generation for products within a product family. Today, principle markets for robot assembly automation (e.g., automobile and consumer electronics) have a relatively short product run rate, ranging from six to twelve months. To accommodate this rapidly changing production demand, in most cases, manufacturers design a series of new products within a product family (i.e., a group of products derived from a common product platform) so that they have similar assembly sequences. Seemingly robot programming for a new product within a product family does not require extensive programming adjustments, but it is still complex and manufacturers spend large amount time and money in programming robots for each new product. To solve this problem, we present a new assembly sequence generation method. With a case-based planner, we generate feasible and similar-to-real assembly sequence candidates for a new product from assembly sequences of old products within a product family. Then, we use geometric information extracted from its CAD model to evaluate those candidate sequences. Unlike those used by previous studies, our method can generate an assembly sequence that is more suitable for the concept of a reconfigurable assembly system for general-purpose robots (assembly sequence generation for robot programming); more similar to assembly sequences used in real assembly systems (case-based planning); and more properly evaluated for robot assembly (a CAD model). We apply the proposed method to a toy airplane assembly scenario and demonstrate its feasibility.

This paper proposes a hybrid feature selection method for predicting user influence on Twitter. A set of candidate features from Twitter is identified based on the five attributes of influencers defined in sociology. Firstly, less relevant features are filtered out with a feature-weighting algorithm. Then the Sequential Backward Floating Selection is utilized as the search strategy. A Back Propagation Neural Network is employed to evaluate the feature subset at each step of searching. Finally, an optimal feature set is obtained for predicting user influence with a high degree of accuracy. Experimental results are provided based on a real world Twitter dataset including seven million tweets associated with 200 popular users. The proposed method can provide a set of features that could be used as a solid foundation for studying complicated user influence evaluation and prediction.

The helicopter fuselage has the characteristics of compact structure space, complex coordination relationship and weak rigidity, etc. It is difficult to control the accuracy of the desired shape. According to the high precision and quality requirement of final assembly for helicopter fuselage, the application status of automatic assembly technology in aviation manufacturing was analyzed. Then the difficulties of automatic assembly technology for fuselage final assembly were also discussed. In this paper, an automatic assembly system for the final assembly of helicopter fuselage was designed, and relevant key technologies of the assembly system mentioned above were proposed. The designed system have the functions of automatic positioning, clamping, measuring, drilling, feeding and blind fastener riveting for fuselage components, which can improve the assembly quality and efficiency of helicopter fuselage during the final assembly process.

Determining disease-related variations of the anatomy and function is an important step in better understanding diseases and developing early diagnostic systems. In particular, image-based multivariate prediction models and the “relevant features” they produce are attracting attention from the community. In this article, we present an empirical study on the relevant features produced by two recently developed discriminative learning algorithms: neighborhood approximation forests (NAF) and the relevance voxel machine (RVoxM). Specifically, we examine whether the sets of features these methods produce are exhaustive; that is whether the features that are not marked as relevant carry disease-related information. We perform experiments on three different problems: image-based regression on a synthetic dataset for which the set of relevant features is known, regression of subject age as well as binary classification of Alzheimer’s Disease (AD) from brain Magnetic Resonance Imaging (MRI) data. Our experiments demonstrate that aging-related and AD-related variations are widespread and the initial sets of relevant features discovered by the methods are not exhaustive. Our findings show that by knocking-out features and re-training models, a much larger set of disease-related features can be identified.KeywordsSynthetic DatasetRelevance Vector MachineCommon Coordinate SystemRandom Permutation TestRelevant PixelThese 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.

The paper proposes a novel Object Shape Error Response (OSER) approach to estimate the dimensional and geometric variation of assembled products and then, relate, these to assembly process parameters, which can be interpreted as root causes (RC) of the object shape defects caused by the assembly system. The OSER approach leverages Bayesian 3D-Convolutional Neural Networks for RC isolation. Compared with the existing methods, the proposed approach (i) addresses a novel problem of applying deep learning for object shape error identification instead of nominal shape identification (object detection); and, (ii) overcome significant limitations in current approaches for Root Cause Analysis (RCA) of assembly systems which are frequently ill-conditioned (collinear fault patterns) with high fault-multiplicity (frequently encountered in 6-sigma strategy) while quantifying the prediction uncertainty. The verification of the proposed methodology is conducted based on the CAE model of a multi-stage assembly process for automotive SUV door made of compliant parts. The results demonstrate performance with R-Squared $(\pmb{R}^{\mathbf{2}})=\mathbf{0.98}$ and Mean Absolute Error $(\mathbf{MAE})=\mathbf{0.05 mm}$ , thus improving RCA capabilities by 29% as compared to state-of-the-art methods.

Polyhedral Representation and Adjacency Graph in n-dimensional Digital Images

Non-Intrusive Load Monitoring (NILM) refers to the analysis of the aggregated current and voltage measurements of Home Electrical Appliances (HEAs) recorded by the house electrical panel. Such methods aim to identify each HEA for a better control of the energy consumption and for future smart grid applications. Here, we are interested in an event-based NILM pipeline, and particularly in the HEAs’ recognition step. This paper focuses on the selection of relevant and understandable features for efficiently discriminating distinct HEAs. Our contributions are manifold. First, we introduce a new publicly available annotated dataset of individual HEAs described by a large set of electrical features computed from current and voltage measurements in steady-state conditions. Second, we investigate through a comparative evaluation a large number of new methods resulting from the combination of different feature selection techniques with several classification algorithms. To this end, we also investigate an original feature selection method based on a deep neural network architecture. Then, through a machine learning framework, we study the benefits of these methods for improving Home Electrical Appliance (HEA) identification in a supervised classification scenario. Finally, we introduce new transfer learning results, which confirm the relevance and the robustness of the selected features learned from our proposed dataset when they are transferred to a larger dataset. As a result, the best investigated methods outperform the previous state-of-the-art results and reach a maximum recognition accuracy above 99% on the PLAID evaluation dataset.