In this paper, we designed and implemented a moving object prediction and grasping system that enables a robot manipulator using a two-finger gripper to grasp moving objects on a conveyor and a circular rotating platform. There are three main parts: (i) moving object recognition, (ii) moving object prediction, and (iii) system realization and verification. In the moving object recognition, we used the instance segmentation algorithm of You Only Look At CoefficienTs (YOLACT) to recognize moving objects. The recognition speed of YOLACT can reach more than 30 fps, which is very suitable for dynamic object recognition. In addition, we designed an object numbering system based on object matching, so that the system can track the target object correctly. In the moving object prediction, we first designed a moving position prediction network based on Long Short-Term Memory (LSTM) and a grasping point prediction network based on Convolutional Neural Network (CNN). Then we combined these two networks and designed two moving object prediction networks, so that they can simultaneously predict the grasping positions and grasping angles of multiple moving objects based on image information. In the system realization and verification, we used Robot Operating System (ROS) to effectively integrate all the programs of the proposed system for the camera image extraction, strategy processing, and robot manipulator and gripper control. A laboratory-made conveyor and a circular rotating platform and four different objects were used to verify that the implemented system could indeed allow the gripper to successfully grasp moving objects on these two different object moving platforms.

No abstract

No abstract

In today’s high-order health examination, imaging examination accounts for a large proportion. Computed tomography (CT), which can detect the whole body, uses X-rays to penetrate the human body to obtain images. Its presentation is a high-resolution black-and-white image composed of gray scales. It is expected to assist doctors in making judgments through deep learning based on the image recognition technology of artificial intelligence. It used CT images to identify the bladder and lesions and then segmented them in the images. The images can achieve high accuracy without using a developer. In this study, the U-Net neural network, commonly used in the medical field, was used to extend the encoder position in combination with the ResBlock in ResNet and the Dense Block in DenseNet, so that the training could maintain the training parameters while reducing the overall identification operation time. The decoder could be used in combination with Attention Gates to suppress the irrelevant areas of the image while paying attention to significant features. Combined with the above algorithm, we proposed a Residual-Dense Attention (RDA) U-Net model, which was used to identify organs and lesions from CT images of abdominal scans. The accuracy (ACC) of using this model for the bladder and its lesions was 96% and 93%, respectively. The values of Intersection over Union (IoU) were 0.9505 and 0.8024, respectively. Average Hausdorff distance (AVGDIST) was as low as 0.02 and 0.12, respectively, and the overall training time was reduced by up to 44% compared with other convolution neural networks.