Search results

This paper aims to quickly obtain an accurate and complete dense three-dimensional map of indoor environment with lower cost, which can be directly used in navigation.

This paper proposes an improved ORB-SLAM2 dense map optimization algorithm. This algorithm consists of three parts: ORB feature extraction based on improved FAST-12, feature point extraction with progressive sample consensus (PROSAC) and the dense ORB-SLAM2 algorithm for mapping. Here, the dense ORB-SLAM2 algorithm adds LoopClose optimization thread and dense point cloud map and octree map construction thread. The dense map is computationally expensive and occupies a large amount of memory. Therefore, the proposed method takes higher efficiency, voxel filtering can reduce the memory while ensuring the density of the map and then use the octree format to store the map to further reduce memory.

The improved ORB-SLAM2 algorithm is compared with the original ORB-SLAM2 algorithm, and the experimental results show that the map through improved ORB-SLAM2 can be directly used in navigation process with higher accuracy, shorter tracking time and smaller memory.

The improved ORB-SLAM2 algorithm can obtain a dense environment map, which ensures the integrity of data. The comparisons of FAST-12 and improved FAST-12, RANSAC and PROSAC prove that the improved FAST-12 and PROSAC both make the feature point extraction process faster and more accurate. Voxel filter helps to take small storage memory and low computation cost, and octree map construction on the dense map can be directly used in navigation.

The purpose of this paper is to put forward a path planning method in complex environments containing dynamic obstacles, which improves the performance of the traditional A* algorithm, this method can plan the optimal path in a short running time.

To plan an optimal path in a complex environment with dynamic and static obstacles, a novel improved A* algorithm is proposed. First, obstacles are identified by GoogLeNet and classified into static obstacles and dynamic obstacles. Second, the ray tracing algorithm is used for static obstacle avoidance, and a dynamic obstacle avoidance waiting rule based on dilate principle is proposed. Third, the proposed improved A* algorithm includes adaptive step size adjustment, evaluation function improvement and path planning with quadratic B-spline smoothing. Finally, the proposed improved A* algorithm is simulated and validated in real-world environments, and it was compared with traditional A* and improved A* algorithms.

The experimental results show that the proposed improved A* algorithm is optimal and takes less execution time compared with traditional A* and improved A* algorithms in a complex dynamic environment.

This paper presents a waiting rule for dynamic obstacle avoidance based on dilate principle. In addition, the proposed improved A* algorithm includes adaptive step adjustment, evaluation function improvement and path smoothing operation with quadratic B-spline. The experimental results show that the proposed improved A* algorithm can get a shorter path length and less running time.

This study aims to develop a data fusion method for powder-bed fusion (PBF) process monitoring based on process image information. The data fusion method can help improve process condition identification performance, which can provide guidance for further PBF process monitoring and control system development.

Design of reliable process monitoring systems is an essential approach to solve PBF built quality. A data fusion framework based on support vector machine (SVM), convolutional neural network (CNN) and Dempster-Shafer (D-S) evidence theory are proposed in the study. The process images which include the information of melt pool, plume and spatters were acquired by a high-speed camera. The features were extracted based on an appropriate image processing method. The three feature vectors corresponding to the three objects, respectively, were used as the inputs of SVM classifiers for process condition identification. Moreover, raw images were also used as the input of a CNN classifier for process condition identification. Then, the information fusion of the three SVM classifiers and the CNN classifier by an improved D-S evidence theory was studied.

The results demonstrate that the sensitivity of information sources is different for different condition identification. The feature fusion based on D-S evidence theory can improve the classification performance, with feature fusion and classifier fusion, the accuracy of condition identification is improved more than 20%.

An improved D-S evidence theory is proposed for PBF process data fusion monitoring, which is promising for the development of reliable PBF process monitoring systems.

The purpose of this paper is to analyze the changes in its importance due to the maintenance and repair of components.

In this paper, a concept of time-varying importance measure is proposed to solve the problem of component importance change caused by maintenance. When the system is broken-down, the probability difference between the component works well after repairing and the component break down before repairing is solved, this difference is measured as an index of time-varying importance method. Then, the approach has been verified by the CNC machine tool.

The paper provides a method to analyze the importance of changes of components in the system due to maintenance. The time-varying importance measure can evaluate the component importance anytime during its whole life span, and it has the ability to find out the most responsible component for a system failure in the actual production process. What is more, it provides guidance for the next maintenance work.

The proposed method can guide the next maintenance time according to the change of component performance caused by each maintenance activity of the manufacturing system, and avoid the waste of resources caused by repeated maintenance.

This paper aims to improve the meshing effect of the gear teeth. It is recommended to analyze the deformation difference between the inner and outer surfaces of the flexspline. The purpose of this paper is to modify the profile of the flexspline based on the deformation difference to improve the transmission accuracy and operating life of the harmonic drive.

In this paper, ring theory is used to calculate the deformation difference of the inner and outer surfaces of the flexspline, and the actual tooth profile of the flexspline is corrected based on the deformation difference. Then, the flexspline is divided into multiple sections along the axial direction, so that the three-dimensional tooth profile of the flexspline is modified to improve the gear tooth meshing effect.

This paper proves the effect of the deformation difference between the inner and outer surfaces of the flexspline on the tooth backlash, which affects the transmission accuracy and life of the harmonic drive. It is recommended to modify the tooth profile of the flexspline based on the deformation difference, so as to ensure the tooth meshing effect.

This paper provides a new way for the optimization of the three-dimensional tooth profile design of the harmonic drive.

The purpose of this paper is to improve the positioning accuracy of 6-Dof serial robot by the way of error compensation and sensitivity analysis.

In this paper, the Denavit–Hartenberg matrix is used to construct the kinematics models of the robot; the effects from individual joint and several joints on the end effector are estimated by simulation. Then, an error model based on joint clearance is proposed so that the positioning accuracy at any position of joints can be predicted for compensation. Through the simulation of the curve path, the validity of the error compensation model is verified. Finally, the experimental results show that the error compensation method can improve the positioning accuracy of a two joint exoskeleton robot by nearly 76.46%.

Through the analysis of joint error sensitivity, it is found that the first three joints, especially joint 2, contribute a lot to the positioning accuracy of the robot, which provides guidance for the accuracy allocation of the robot. In addition, this paper creatively puts forward the error model based on joint clearance, and the error compensation method which decouples the positioning accuracy into joint errors.

It provides a new idea for error modeling and error compensation of 6-Dof serial robot. Combining sensitivity analysis results with error compensation can effectively improve the positioning accuracy of the robot, and provide convenience for welding robot and other robots that need high positioning accuracy.

Workpiece sorting is a key link in industrial production lines. The vision-based workpiece sorting system is non-contact and widely applicable. The detection and recognition of workpieces are the key technologies of the workpiece sorting system. To introduce deep learning algorithms into workpiece detection and improve detection accuracy, this paper aims to propose a workpiece detection algorithm based on the single-shot multi-box detector (SSD).

Propose a multi-feature fused SSD network for fast workpiece detection. First, the multi-view CAD rendering images of the workpiece are used as deep learning data sets. Second, the visual geometry group network was trained for workpiece recognition to identify the category of the workpiece. Third, this study designs a multi-level feature fusion method to improve the detection accuracy of SSD (especially for small objects); specifically, a feature fusion module is added, which uses “element-wise sum” and “concatenation operation” to combine the information of shallow features and deep features.

Experimental results show that the actual workpiece detection accuracy of the method can reach 96% and the speed can reach 41 frames per second. Compared with the original SSD, the method improves the accuracy by 7% and improves the detection performance of small objects.

This paper innovatively introduces the SSD detection algorithm into workpiece detection in industrial scenarios and improves it. A feature fusion module has been added to combine the information of shallow features and deep features. The multi-feature fused SSD network proves the feasibility and practicality of introducing deep learning algorithms into workpiece sorting.

The purpose of this paper is to identify the critical components of a complex system by using survival signature. First, a complex system is abstracted with varying scales and generates a multi-levels model. Then reliability evaluations can be conducted by survival signature from rough to fine for tracing and identifying them. Finally, the feasibility of the proposed approach is demonstrated by an actual production system.

The paper mainly applies a multi-level evaluating strategy for the reliability analysis of complex systems with components of multiple types. In addition, a multi-levels model of a complex system is constructed and survival signature also used for evaluation.

The proposed approach was demonstrated to be the feasibility by an actual production system that is used in the case study.

The case study was performed on a system with simple network structure, but the proposed approach could be applied to systems with complex ones. However, the approach to generate the digraphs of abstraction levels for complex system has to be developed.

So far the approach has been used for the reliability analysis of a machining system. The approach that is proposed for the identification of critical components also can be applied to make maintenance decision.

The multi-level evaluating strategy that was proposed for reliability analysis and the identification of critical components of complex systems was a novel method, and it also can be applied as index to make maintenance planning.

Feature selection is an essential step in data mining. The core of it is to analyze and quantize the relevancy and redundancy between the features and the classes. In CFR feature selection method, they rarely consider which feature to choose if two or more features have the same value using evaluation criterion. In order to address this problem, the standard deviation is employed to adjust the importance between relevancy and redundancy. Based on this idea, a novel feature selection method named as Feature Selection Based on Weighted Conditional Mutual Information (WCFR) is introduced. Experimental results on ten datasets show that our proposed method has higher classification accuracy.

Reverse engineering methodology paves an efficient way for simulating and manufacturing existing objects with complex shapes, and a range of applications also has shown this approach to be feasible and efficient. However, in many applications, a lot of sub‐tasks in reverse engineering are usually not done in the same place, they need to be done cooperatively over the Internet or Intranet. So, this paper is concerned with developing a novel e‐service platform for remote service in reverse engineering applications based on mobile agent technology, and some correlative enabling technologies such as data compression, security considerations, agent models and so on. The framework was programmed using the Java Sevlets and Java Beans component models, and data transferring and processing were implemented based on the dispatch/retract mechanisms of mobile agents. The mobile agent is developed with IBM's Aglets Workbench, and the feasibility of the proposed method has been verified by a case.