The international standard ISO 9283:1998 is popular for performance tests of industrial robots at present. It is desirable that the tests described in this standard should be sensitive to error sources of robot end positioning/orientation. In this paper, first, the kinematic and the joint stiffness model parameters are identified experimentally for two models of six-DOF (degree-of-freedom) serial industrial robots (i.e., the ABB IRB 1410 and UR5 robots). Then, the standard deviations of the derived model parameters are obtained as error inputs for the sensitivity analysis of the performance tests including the positioning/orientation accuracy/repeatability tests. By simulating the error sensitivity of the positioning/orientation accuracy/repeatability test methods for industrial robots, it is analyzed whether the tests described in the ISO 9283:1998 Standard are sensitive to the focused error sources, showing the limitations of the evaluation index of the ISO 9283:1998 Standard. The results show that for six-DOF serial industrial robots, the positioning accuracy test is the key to determining their motion performance. The orientation accuracy and repeatability tests are not necessary if the positioning accuracy and repeatability tests can be done for six-DOF serial industrial robots. Finally, the improvement suggestion of the performance test method is proposed. The research of this paper is beneficial for improving the performance evaluation methods of industrial robots. It can also help robot manufacturing enterprises analyze and improve the positioning/orientation accuracy/repeatability of their products.

In recent years, there have been significant breakthroughs in the field of detection using deep learning technology. What used to be challenging defects for traditional visual methods can now be addressed with the help of deep learning techniques. This paper employs the YOLOv5 network architecture to achieve rapid and precise detection of concrete surface cracks. Additionally, it integrates the C2f module to overcome issues like gradient vanishing and information bottlenecks, which can limit the performance of the traditional YOLOv5 network. The C2f module enhances feature propagation and utilization by introducing strategies such as feature reuse, cross-stage partial connections, and attention mechanisms, thereby improving feature representation and information flow. Various training techniques are also applied to enhance training speed and detection accuracy, including weighted residual connections (WRC), cross-stage partial connections (SCP), cross mini-batch normalization (CmBN), self-adaptive training (SAT), and mish activation function. As a result, the system achieves a detection accuracy of 96.91% on a concrete crack detection dataset.

This paper proposes an effective traffic light identification scheme based on TensorFlow API and HSV color space. The program uses a distortion-corrected wide-angle camera to collect the road images in real time and utilizes the RCNN target detection interface, which is provided by the Google TensorFlow framework, to track the precise bounding box of the traffic light in the image. The obtained corner information is recorded for cropping ROI image from the original. Then using Open CV to smooth the ROI image and enhance the contrast. After converting the ROI image from BGR to HSV color space, the result of the traffic light can be consulted on the H channel according to the area of the connected domain.

Purpose: The purpose of this paper is to propose an availability modeling method of complex multiple units system (CMUS) based on the multi-agent technique. Design/methodology/approach: Based on the multi-agent technique, this paper describes the availability model structure for CMUS and develops agent-based models of components, maintenance policies, maintenance tools, maintenance fields, and maintenance staff, as well as the communication method among the different agents. On the basis of the agent-based availability modeling theory, the availability simulation scheme of CMUS is given using MATLAB. Thus, the availability modeling theory of CMUS and its simulation method are developed. To demonstrate the applicability of the proposed availability modeling method, a numerical example is given. Findings: The proposed agent-based modeling method is applicable to availability modeling of CMUS, including the modeling of component failure, maintenance tools/fields/staff, maintenance policy, and structural/economic dependence among components. Practical implications: As a bottom-top, modular, expandable, and reusable modeling theory, the agent-based modeling method might be useful for availability modeling of different CMUSs in reality. Originality/value: The multi-agent technique is introduced into availability modeling of multi-component systems in this paper. Thus, it is possible to model failure of many components, maintenance policies, maintenance tools, maintenance fields, and maintenance staff together for availability analysis of complex systems of equipment.

Person re-identification aims to identify the same pedestrians captured by various cameras from different viewpoints in multiple scenarios. Occlusion is the toughest problem for practical applications. In video-based ReID tasks, motion information can be easily obtained from sampled frames, and provide discriminative human part representations. However, most motion-based methodologies are designed for video frames which are not suitable for processing single static image input. In this paper, we propose a Motion-Aware Fusion (MAF) network, aiming to acquire motion information from static images in order to improve the performance of ReID tasks. Specifically, a visual adapter is introduced to enable visual feature extraction, either from image or video data. We design a motion consistency task to guide the motion-aware transformer to learn representative human-part motion information and greatly improve the learning quality of features of occluded pedestrians. Extensive experiments on popular holistic, occluded, and video datasets demonstrate the effectiveness of our proposed method. This method outperforms state-of-the-art approaches by improving the mean average precision (mAP) by 1.5% and rank-1 accuracy by 1.2% on the challenging Occluded-REID dataset. At the same time, it surpasses other methods on the MARS dataset with an improvement of 0.2% in mAP and 0.1% in rank-1 accuracy.

This paper proposes a vision-based online fruit and vegetable inspection system. This system consists of two parts: the online detection of fruits and vegetables and the online weighing measurement of fruits and vegetables. Online detection system can detect the species of fruits and vegetables, and through the database to get the unit price of fruits and vegetables. The image processing includes image pre-processing, filtering, background removal, noise filtering and hole filling for recognition, and analysis of blocks. The weight of the fruits and vegetables is obtained by gravity sensor. Prices of fruits and vegetables can be displayed in real time in order to achieve the goal of unattended retail. This study built an online fruits and vegetables detection system which may be accelerated in the future for commercial uses.