Shenyang Institute of Automation,Chinese Academy Of Sciences
Not a member yet
23582 research outputs found
Sort by
Lightweight Real-time Apple Detection Method Based on Improved YOLO v4
No full text针对苹果采摘机器人识别算法包含复杂的网络结构和庞大的参数体量，严重限制检测模型的响应速度问题，本文基于嵌入式平台，以YOLO v4作为基础框架提出一种轻量化苹果实时检测方法（YOLO v4-CA）。该方法使用MobileNet v3作为特征提取网络，并在特征融合网络中引入深度可分离卷积，降低网络计算复杂度；同时，为弥补模型简化带来的精度损失，在网络关键位置引入坐标注意力机制，强化目标关注以提高密集目标检测以及抗背景干扰能力。在此基础上，针对苹果数据集样本量小的问题，提出一种跨域迁移与域内迁移相结合的学习策略，提高模型泛化能力。试验结果表明，改进后模型的平均检测精度为92.23%，在嵌入式平台上的检测速度为15.11f/s，约为改进前模型的3倍。相较于SSD与Faster R-CNN，平均检测精度分别提高0.91、2.02个百分点，在嵌入式平台上的检测速度分别约为SSD和Faster R-CNN的1.75倍和12倍；相较于两种轻量级目标检测算法DY3TNet与YOLO v5s，平均检测精度分别提高7.33、7.73个百分点。因此，改进后的模型能够高效实时地对复杂果园环境中的苹果进行检测，适宜在嵌入式系统上部署，可以为苹果采摘机器人的识别系统提供新的解决思路。</p
Enhanced Terahertz Phase Retrieval Imaging by Unequal Spaced Measurement
No full textTerahertz lensless phase retrieval imaging is a promising technique for non-destructive inspection applications. In the conventional multiple-plane phase retrieval method, the convergence speed due to wave propagations and measures with equal interval distance is slow and leads to stagnation. To address this drawback, we propose a nonlinear unequal spaced measurement scheme in which the interval space between adjacent measurement planes is gradually increasing, it can significantly increase the diversity of the intensity with a smaller number of required images. Both the simulation and experimental results demonstrate that our method enables quantitative phase and amplitude imaging with a faster speed and better image quality, while also being computationally efficient and robust to noise.</p
基于Mobile Net-SSD算法的铁矿石品位识别
No full text矿石的品位是指导矿山生产的必要指标。为实现对铁矿石品位的智能化识别以及模型在便携设备中的搭载，对不同铁矿石的不同品位进行了数据增强处理，利用MobileNet搭建SSD神经网络作为方法的第一个判断模块，在训练神经过程中采用迁移学习思想以及早停法对训练进行加速，进而生成铁矿石品位识别模型。测试集验证结果表明，模型对于矿石品位图像识别正确率大于97%，模型可以很好地区分不同品位的铁矿石，可以为其他矿山品位识别提供参考。</p
Fast and Accurate Hand Visual Detection by Using a Spatial-Channel Attention SSD for Hand-Based Space Robot Teleoperation
No full textSpace robot teleoperation is an important technology in the space human-robot interaction and collaboration. Hand-based visual teleoperation can make the operation more natural and convenient. The fast and accuracy hand detection is one of the most difficult and important problem in the hand-based space robot teleoperation. In this work, we propose a fast and accurate hand detection method by using a spatial-channel attention single shot multibox detector (SCA-SSD). The SSD framework is used and improved in our method by introducing spatial-channel attentions with feature fusion. To increase the restricted receptive field in shallow layers, two shallow layers are fused with deep layers by using feature fusion modules. And spatial attention and channel-wise attention are also used to extract more efficient features. This method can not only ease the computational burden but also bring more contextual information. To evaluate the effectiveness of the proposed method, experiments on some public datasets and a custom astronaut hand detection dataset (AHD) are conducted. The results show that our method can improve the hand detection accuracy by 2.7% compared with the original SSD with only 15 fps speed drops. In addition, the space robot teleoperation experiment proves that our hand detection method can be well utilized in the space robot teleoperation system.</p
Incremental Learning and Fault-tolerant Classifier for Myoelectric Pattern Recognition against Multiple Bursting Interferences
No full textBursting interference that causes a sudden and significant change in surface electromyography (sEMG) characteristics, can reduce the stability and security of myoelectric assistive robots. Current adaptation strategies for progressive-generated interference are incapable of dealing with bursting interference. To address this problem, an incremental learning and fault-tolerant classifier (ILFTC) was proposed by combining a Gaussian mixture model (GMM) ensemble and linear discriminant analysis (LDA), in conjunction with online update and marginalization schemes. Subsequently, an ILFTC-based myoelectric pattern recognition (MPR) strategy was developed to improve the robustness of MPR against multiple interferences, including outlier motion and missing/fault data owing to electrode loosening. Experiments on hand/wrist motions were conducted to validate the anti-interference performance of the ILFTC. Experimental results showed that the ILFTC could effectively resist the two types of bursting interference and produce a significant improvement in the recognition performance over traditional classifiers, as well as the methods presented in previous studies. The results show that the proposed method has the potential to enhance the robustness of myoelectric assistive robots.</p
Elastic deformation modeling of series robots with consideration of gravity
No full textIn this paper, an elastic deformation modeling method of series robots with consideration of gravity that combines the finite element structure method (FESM) with the virtual joint method (VJM) is proposed to improve the positioning accuracy of robots. This method has characteristics of low computational complexity, high precision, as well as high real time. Compared with the previous research, the influence of joint and link mass on the elastic deformation of robots can be considered. Firstly, the entire robot model is split into several independent components and these components are expressed as a combination of a rigid body and a 6-Dofs virtual joint. Through the VJM, an extended kinematic model of series robots is built with these 6-Dofs virtual joints. Secondly, the stiffness parameters of robot components or virtual joints are extracted and considered comprehensively by the FESM. Thirdly, according to the extended kinematic model, the elastic deformation model is established to obtain the positioning error caused by external wrenches and gravity of robots. Lastly, a series robot SHIR5-II is taken as an illustration to perform elastic deformation modeling by the proposed method. Based on laser tracker and finite element software, the static compliance test, the static compliance simulation and the 7-Dofs VJM-based elastic deformation modeling of SHIR5-II robot are performed to verify the effectiveness of this modeling method. © 2022, The Author(s), under exclusive licence to Springer-Verlag GmbH Germany, part of Springer Nature.</p
A Quality-Related Fault Detection Method Based on the Dynamic Data-Driven Algorithm for Industrial Systems
No full textFor nearly a decade, quality-related fault detection algorithms have been widely used in industrial systems. However, the majority of these detection strategies rely on static assumptions of the operating environment. In this paper, taking the time series of variables into consideration, a dynamic kernel entropy component regression (DKECR) framework is proposed to address the instability of quality-related fault detection due to the existing dynamic characteristics. Compared with the typical kernel entropy component analysis method, the proposed method constructs the relationship between process states and quality states to further interpret the direct effect on the product taken by the fault. In the proposed approach, process measurements are converted to a lower-dimensional subspace with a specific angular structure that is more comprehensive than traditional subspace approaches. In addition, the angular statistics and their relevant thresholds are exploited to enhance the quality-related fault detection performance. Finally, the proposed method will be compared with three methods by means of a numerical example and two industrial scenarios to demonstrate its practicality and effectiveness
3D interest point detection using balance-distortion oriented selection
No full textInterest point detection is a challenging problem in 3D objects. Compared to traditional corner detection based on the curvature, this paper proposes a novel method that quantifies the balance and uniformity of local geometric structures based on the distribution of vertex neighborhoods. We first define the neighborhoods of vertices and structure them within the two-ring, instead of constructing the overall mesh, so as to avoid the interference between the neighborhoods of different vertices. Then we introduce the concept "balance-distortion" to describe the geometric features of the local structure. The experimental results show that the proposed algorithm is robust against noise and invariant to geometric transformation. In addition, compared with the corner detection, more feature points that do not satisfy the balance and direction uniformity are detected, and the distribution of interest point is more uniform
Publisher Correction: Optomechanical dissipative solitons (Nature, (2021), 600, 7887, (75-80), 10.1038/s41586-021-04012-1)
No full textIn the version of this article initially published, there were errors in Fig. 2h, i. The y-axis labels for both panels, now reading (Formula presented.), were missing minus signs after “Ωm.” The changes have been made to the online version of the article.</p
一种面向大质量大长径比构件的定位工装
No full text本发明涉及定位工装，具体地说是一种面向大质量大长径比构件的定位工装，包括可移动式主体托盘部分、分体式卡箍部分及多个可调节式拉动连杆部分，每个可调节式拉动连杆部分的上端分别与分体式卡箍部分铰接，每个可调节式拉动连杆部分的下端分别铰接于可移动式主体托盘部分，每个可调节式拉动连杆部分的内侧均设有安装于可移动式主体托盘部分上的定位整体座组件。本发明适应产品种类广泛，调节范围广，定位牢固，运动稳定，特点突出，工作可靠