White blood cell detection plays an integral role in diagnosing pathologies such as leukemia and gestational diabetes. Despite this, conventional image-based white blood cell classification ...methodologies encounter obstacles including inaccurate cell segmentation and labor-intensive artificial feature extraction. Contrarily, Convolutional Neural Networks (CNNs) have the capacity to learn features autonomously from raw images, thereby offering a novel and effective solution for blood cell detection. Notwithstanding, the features ascertained by a solitary CNN tend to be unidirectional. Conversely, ensemble learning combines results from numerous networks, thus ensuring an adequate acquisition of feature information and subsequently enhancing the model's overall efficacy. Consequently, this study introduces a method for white blood cell classification underpinned by ensemble CNNs. Initially, three high-performing CNNs possessing disparate structures, namely VGG16, ResNet50, and Inception V3, are enlisted as base learners to augment the diversity of base learners. Subsequently, the Gompertz function is employed to strategize the ensemble learning combination strategy, taking into consideration the prediction confidence and fuzzy level of each base learner. Ultimately, the ensemble CNN model is developed, incorporating learning outcomes from several singular models and utilizing diversified information to achieve white blood cell classification. Empirical results indicate that the ensemble learning technique advanced in this study enables accurate and reliable white blood cell classification, demonstrating potential clinical value.
Particle swarm optimization (PSO) is a widely used method that can provide good parameters for the motion controller of mobile robots. In this paper, an improved PSO algorithm that optimize the ...control PID parameters of a specific robot have been proposed. This paper first presents a brief review of recently proposed PSO methods, and then presents a detailed analysis of the PID optimization algorithm, which uses H∞ theory to reduce the search space and fuses the information entropy to ensure the diversity of particles. Simulations in Matlab show that the algorithm can improve the convergence speed and get a better global optimization ability than the standard PSO algorithm. Experimental results present a sound effects for the control of the negative pressure adsorption motor in the power grid pipeline robot during its adsorption along the circular movements, which verifies the effectiveness of the proposed method.
Abstract
In order to enable a robot to do the omni-directional mobility for inspection in the pipeline of Gas Insulation Switchgear, a negative pressure adsorption pipeline robot is designed. ...Firstly, the mechanical structure was designed for the omni-directional mobility based on the negative pressure adsorption. The internal surface motion model of the pipeline robot was built, and a variable adsorption control method was proposed by mechanical analysis. Simulations and experiments showed that the robot could steadily adsorb on any position of the pipeline, which validated the capability of omni-directional mobility for the designed robot. This robot can be used for inspection of the Gas Insulation Switchgear in electric power industry in the near future.
A solving method for the inverse-kinematics (IK) problem of a kind of 6 degree of freedom (DOF) manipulator with the axes of the 3 adjacent rotational joints in the arm part,shoulder,elbow and wrist ...parallel is presented. The merit of this method is that,by utilizing the geometrical relation between the end-effector,hand and wrist of the manipulator determined by its structure,the position of the wrist corresponding to a desired end-effector orientation and position can be solved at the first step in the whole calculation process. And then,the solution of a 6-variable equations,which describes the relationship between the 6 joint positions of the manipulator and its end-effector orientation and position is reduced into two 3-variable equations,one of which describes the relationship between the wrist position and the positions of 3 joints distributed in the shoulder and elbow,and the other describes the relationship between the end-effector orientation and the positions of the rest 3 joint distributed in the
Abstract
Robotic operation is an effective way to upgrade the live-line maintenance safety, efficiency, and quality. This paper proposes an impedance control-based teleoperation system to improve the ...adaptability of live-line maintenance robot in outdoors environment. The key technology of this system is utilizing three virtual spring-damper systems to model the elastic connection between the end-effector of slave manipulator and the environment, between the end-effector of slave manipulator and the counterpart of master haptic device, and between the end-effector of master haptic device and its base, respectively. Experiment results show that, under control of our proposed teleoperation system, the slave manipulator is able to track the motion of master haptic device and the robot is able to complete a set of complex action to peel the coat off the cable.
Traditional white blood cell detection usually requires artificial extraction of cell features, which have higher resolution and contain more detailed information. However, due to the complex ...background of blood cell images and the large individual differences between cells, artificial feature extraction methods have poor generalization for different image data sets. Convolutional neural networks can extract deep features with stronger semantic information through their powerful self-learning capabilities. However, the pooling layer will cause the loss of some detailed information, while also ignoring the relationship between the whole and the part. Therefore, this paper innovatively proposes a white blood cell classification algorithm that combines deep learning features with artificial features. This algorithm not only uses artificial features, but also combines the self-learning capabilities of Inception V3 to make full use of the feature information of the image. At the same time, this paper introduces the transfer learning method to solve the problem of the dataset limitation. The final experimental results show that by fusing deep learning features with artificial features, the classification accuracy of white blood cell images reaches more than 99%. The proposed method has both low complexity and high accuracy, which makes it of great reference value for other medical image detection problems.
Visual SLAM has rich application scenarios and can work both indoors and outdoors. However, it is greatly affected by ambient light and cannot work in dark places. There are cumulative errors in map ...construction, which poses challenging requirements for precise positioning and low computing costs. To solve this problem, this paper introduces a SLAM system based on fusion of vision and inertial measurement unit (IMU). Aiming at the angle error existing in the raw data of the IMU sensor, a mean filtering algorithm is designed in this paper. The IMU data is filtered and the visual information is used as input to the SLAM system to improve the positioning accuracy and robustness of the system. Experimental results prove that the mean filtering algorithm for IMU data processing can effectively decrease the angular error of the system. The implementation of the proposed method shows its potential in practical applications, especially in environments with high precision requirements.
Tracking the actions of vehicles at crossroads and planning safe trajectories will be an effective method to reduce the rate of traffic accident at intersections. It is to resolve the problem of the ...abrupt change because of the existence of drivers' voluntary choices. In this paper, we make approach of an improved IMM tracking method based on trajectory generation, abstracted by trajectory generation algorithm, to improve this situation. Because of the similarity between human-driving trajectory and programming trajectory which is generated by trajectory-generated algorithm, the improved IMM method performs well in tracking moving vehicles with some sudden changes of its movement. A set of data is collected for experiments when an object vehicle takes a sudden left turn in intersection scenario. To compare the experiment results between IMM method with trajectory generation model and the one without, tracking error of the former decreases by 75% in particular scenario.
This paper present a power lines automatic extraction method from airborne LiDAR point cloud. Firstly, ground points are removed by automatic filtering method based on fluctuant feature of terrain. ...Vegetation points, building points and part pylon points are also removed by dimensionality feature and direction feature. Secondly, 2D Hough transform and least square fitting are used to fit center line equation of power lines. And then, the laser point of each power lines can be obtained by center line equation. In this step, power lines projection overlap in the horizontal plane are considered as well. Finally, block centroid calculation method is used to calculate 3D nodes of each power lines. These nodes are used to output the power lines vector. The experimental result shows that the proposed method can extract complete power lines from airborne LiDAR point cloud. This method has some practical meaning for power line inspection.