基于自适应神经网络的磁编码器角度误差补偿算法

针对磁编码器中各类误差导致的解码精度低的问题,在神经网络原理的基础上提出单层自适应神经网络对正余弦信号中存在的幅值不相等、相位不正交、直流偏置、谐波与噪声等误差进行补偿。采用锁相环算法对补偿后的正余弦信号进行解码。电路中采用TLE5501磁阻芯片检测角度变化,利用TL082C运放芯片对信号进行调理,最后采用STM32G431单片机验证算法的性能。通过仿真与试验验证,证明了该算法的有效性与可行性。     

AR-based deep learning for real-time inspection of cable brackets in aircraft

In the process of aircraft assembly, there exist numerous and ubiquitous cable brackets that shall be installed on frames and subsequently need to be manually verified with CAD models. Such a task is usually performed by special operators, hence is time-consuming, labor-intensive, and error-prone. In order to save the inspection time and increase the reliability of results, many researchers attempt to develop intelligent inspection systems using robotic, AR, or AI technologies. However, there is no comprehensive method to achieve enough portability, intelligence, efficiency, and accuracy while providing intuitive task assistance for inspectors in real time. In this paper, a combined AR+AI system is introduced to assist brackets inspection in a more intelligent yet efficient manner. Especially, AR-based Mask R-CNN is proposed by skillfully integrating markerless AR into deep learning-based instance segmentation to generate more accurate and fewer region proposals, and thus alleviates the computation load of the deep learning program. Based on this, brackets segmentation can be performed robustly and efficiently on mobile devices such as smartphones or tablets. By using the proposed system, CAD model checking can be automatically performed between the segmented physical brackets and the corresponding virtual brackets rendered by AR in real time. Furthermore, the inspection results can be directly projected on the corresponding physical brackets for the convenience of maintenance. To verify the feasibility of the proposed method, experiments are carried out on a full-scale mock-up of C919 aircraft main landing gear cabin. The experimental results indicate that the inspection accuracy is up to 97.1%. Finally, the system has been deployed in the real C919 aircraft final-assembly workshop. The preliminary evaluation reveals that the proposed real-time AR-assisted intelligent inspection approach is effective and promising for large-scale industrial applications.     

Improved genetic algorithm based on multi-layer encoding approach for integrated process planning and scheduling problem

Integrated process planning and scheduling (IPPS) is of great significance for modern manufacturing enterprises to achieve high efficiency in manufacturing and maximize resource utilization. In this paper, the integration strategy and solution method of IPPS problem are deeply studied, and an improved genetic algorithm based on multi-layer encoding (IGA-ML) is proposed to solve the IPPS problem. Firstly, considering the interaction ability between the two subsystems and the multi-flexibility characteristics of the IPPS problem, a new multi-layer integrated encoding method is designed. The encoding method includes feature layer, operation layer, machine layer and scheduling layer, which respectively correspond to the four sub-problems of IPPS problem, which provides a premise for a more flexible and deeper exploration in the solution space. Then, based on the coupling characteristics of process planning and shop scheduling, six evolutionary operators are designed to change the four-layer coding interdependently and independently. Two crossover operators change the population coding in the unit of jobs, and search the solution space globally. The four mutation operators change the population coding in the unit of gene and search the solution space locally. The six operators are used in series and iteratively optimized to ensure a fine balance between the global exploration ability and the local exploitation ability of the algorithm. Finally, performance of IGA-ML is verified by testing on 44 examples of 14 benchmarks. The experimental results show that the proposed algorithm can find better solutions (better than the optimal solutions found so far) on some problems, and it is an effective method to solve the IPPS problem with the maximum completion time as the optimization goal.     

Difficulty and complexity definitions for assembly task allocation and assignment in human–robot collaborations: A review

This paper presents a literature review on the different aspects of task allocation and assignment problems in human–robot collaboration (HRC) tasks in industrial assembly environments. In future advanced industrial environments, robots and humans are expected to share the same workspace and collaborate to efficiently achieve shared goals. Difficulty- and complexity-aware HRC assembly is necessary for human-centric manufacturing, which is a goal of Industry 5.0. Therefore, the objective of this study is to clarify the definitions of difficulty and complexity used to encourage effective collaboration between humans and robots to leverage the adaptability of humans and the autonomy of robots. To achieve this goal, a systematic review of the following relevant databases for computer science was performed: IEEE Xplore, ScienceDirect, SpringerLink, ACM Digital Library, and ASME Digital Collection. The results extracted from 74 peer-reviewed research articles published until July 2022 were summarized and categorized into four taxonomies for 145 difficulty and complexity definitions from the perspectives of (1) definition-use objectives, (2) evaluation objectives, (3) evaluation factors, and (4) evaluation variables. Next, existing definitions were primarily classified according to the following two criteria to identify potential future studies on the formulation of new definitions for human-centric manufacturing: (1) agent specificity and (2) common aspects in manual and robotic assemblies.     

Differential geometry modeling and application of roller pose and trajectory of robot roller hemming for complex curved surface-curved edge panels

Roller hemming is a relatively new process used to achieve high-precision assembly of auto-body enclosure panels. During the process of roller hemming, accuracy of the roller pose and trajectory affects the hemming quality of the product. The traditional passive method based on robot teaching to determine the pose of the roller is inefficient and time-consuming. In these studies, we proposed an active method for solving roller pose and trajectory based on differential geometry for curved surface-curved edge geometric characteristics of auto-body enclosure panels and multi-pass reciprocating motions of the roller. Firstly, the local coordinate system of the die was constructed based on the Frenet Frame according to the normal vector of the surface of die and the tangent vector of the curved die edge. Secondly, the coordinate system of the die, diameter of the roller, TCP-RTP value, and inclination of the roller were combined to form the roller pose based on a homogeneous transformation matrix. Based on the obtained trajectory curve of the roller reference point, the equal chord deviation error method was used to analyze the roller trajectory. Finally, a roller pose and trajectory solving algorithm was developed based and implemented using PYTHON to obtain the positions and poses of the roller at several discrete reference points. ABAQUS software was subsequently utilized to complete modeling of the roller pose and trajectory. This research supports the multi-field mechanical simulation of robot roller hemming for curved surface-curved edge panels and provides support for determining roller pose and kinematic trajectory of industrial robot roller hemming for curved surface-curved edge panels.     

Adaptive adversarial prototyping network for few-shot prototypical translation

Translating multiple real-world source images to a single prototypical image is a challenging problem. Notably, these source images belong to unseen categories that did not exist during model training. We address this problem by proposing an adaptive adversarial prototype network (AAPN) and enhancing existing one-shot classification techniques. To overcome the limitations that traditional works cannot extract samples from novel categories, our method tends to solve the image translation task of unseen categories through a meta-learner. We train the model in an adversarial learning manner and introduce a style encoder to guide the model with an initial target style. The encoded style latent code enhances the performance of the network with conditional target style images. The AAPN outperforms the state-of-the-art methods in one-shot classification of brand logo dataset and achieves the competitive accuracy in the traffic sign dataset. Additionally, our model improves the visual quality of the reconstructed prototypes in unseen categories. Based on the qualitative and quantitative analysis, the effectiveness of our model for few-shot classification and generation is demonstrated.     

Multi-layer and Multi-scale feature aggregation for DIBR-Synthesized image quality assessment

Depth-Image-Based Rendering (DIBR) is one of the main fundamental techniques for generating new viewpoints in 3D video applications such as multi-viewpoint video (MVV), free viewpoint video (FVV) and virtual reality (VR). Due to the imperfections of color images, depth maps or texture restoration techniques, several types of distortions occur in synthesized views. However, most of related works evaluated the quality of DIBR-synthesized views by only detecting a specific type of distortion, such as stretching, black holes, blurring, etc., which were unable to accurately evaluate the quality of DIBR-synthesized views. In this paper, a new no-reference image quality assessment method is proposed to evaluate the quality of DIBR-synthesized images by combining multi-layer and multi-scale features of images. To be specific, the distortions introduced by different stages of virtual viewpoint synthesis are first analyzed, and then multi-layer and multi-scale features are extracted to estimate the degree of texture and structure distortions. As a result, individual quality scores associated with two types of distortions (e.g., structural distortion and texture distortion) are aggregated to an overall image quality. Experimental results on two publicly available DIBR datasets show that the method has better performance than the state-of-the-art models.Index Terms: image quality assessment, DIBR-synthesized image, distortion correction, BIQA.     

基于张量鲁棒主成分分析的视网膜眼底图像序列变化检测

在计算机辅助诊断系统中,视网膜眼底图像序列的变化检测是一项重要且具有挑战性的任务。针对视网膜眼底图像序列采样帧少、光照干扰大、难以获得稳健的背景模型,提出了一种基于张量鲁棒主成分分析(tensor robust principal component analysis, TRPCA)的变化检测方法。该方法以TRPCA为模型,通过对序列背景扩充,再利用张量分解而获得变化区域：首先,先选择出序列中最接近正常状态的一张图像作为背景模型;然后,通过预处理将单帧背景模型扩张成多帧背景使得背景模型包含更丰富的光照变化;接着,将整个序列建模为一个3维张量体;最后,利用总变分约束背景模型和变化区域的时空连续性,通过Tucker分解分离出背景模型,获得变化区域。实验结果表明,与基于矩阵的鲁棒主成分分析(matrix robust principal component analysis, Matrix RPCA)方法,Masked-RPCA方法以及不加总变分约束的TRPCA方法相比,基于总变分约束的TRPCA方法能够更准确地分离出变化区域,对血管和光照干扰更具有鲁棒性。     

基于差分进化的置换混叠图像盲分离

针对1类存在模糊差异的置换混叠图像,提出了1种基于差分进化的置换混叠图像盲分离算法。选择空域作为特征域,将置换混叠图像的梯度图像分成小块,并为每个小块设定个阈值,组成1个阈值向量。利用差分进化算法获得最优的阈值向量,将置换混叠图像阈值化为二值图像,利用图像形态学实现二值图像分离出置换图像。实验结果表明,与阈值法相比较,本算法能较好地把置换图像从置换混叠图像中分离出来,而不受置换图像位置、大小和个数的限制。