RESUMO
Automated inspection has proven to be the most effective approach to maintaining quality in industrial-scale manufacturing. This study employed the eye-in-hand architecture in conjunction with deep learning and convolutional neural networks to automate the detection of defects in forged aluminum rims for electric vehicles. RobotStudio software was used to simulate the environment and path trajectory for a camera installed on an ABB robot arm to capture 3D images of the rims. Four types of surface defects were examined: (1) dirt spots, (2) paint stains, (3) scratches, and (4) dents. Generative adversarial network (GAN) and deep convolutional generative adversarial networks (DCGAN) were used to generate additional images to expand the depth of the training dataset. We also developed a graphical user interface and software system to mark patterns associated with defects in the images. The defect detection algorithm based on YOLO algorithms made it possible to obtain results more quickly and with higher mean average precision (mAP) than that of existing methods. Experiment results demonstrated the accuracy and efficiency of the proposed system. Our developed system has been shown to be a helpful rim defective detection system for industrial applications.
Assuntos
Aprendizado Profundo , Robótica , Algoritmos , Redes Neurais de ComputaçãoRESUMO
Sorting Municipal Solid Waste (MSW) has helped promote the awareness of sustainable development of environment. A robot equipped with an intelligent deep learning (DL) detection algorithm have been proposed to improve the sorting task. But most of the related studies aimed to better the DL algorithms on MSW detection, and few studies integrated the DL algorithms with a robot to identify the dominated factors to Intelligent MSW Sorter (IMSWS). Therefore, this study is to develop IMSWS prototype to better sort MSW, based on the pick-and-place process, and preliminarily evaluate the dominated factors. First, the delta robot prototype was manufactured, and IMSWS was performed with a camera to acquire the RGB image and the height of a MSW in the conveyor belt. The DL algorithm, YOLOv3 or YOLOv4, detected the type and plane location of the MSWs in the conveyor belt. Next, the sequence program transferred the valid MSW data to the delta robot. After the calculation of the absorbed location of the target MSW was made, the arm of this delta robot moved to absorb and then transfer the MSW to the bin. Results showed that the IMSWS prototype could sort the multi-object MSWs in the MSW stream. Both YOLOv3 and YOLOv4 reached high detection accuracy on the MSW image dataset. However, the improvement should be made in the actually moving MSW stream even though the YOLOv4 performed the acceptable detection accuracy. The gripping stability of the arm mainly dominated the performance of IMSWS, and this should be improved first.