Defective shafts need to be classified because some defective shafts can be reworked to avoid replacement costs. Therefore, the detection and classification of shaft surface defects has important engineering application value. However, in the factory, shaft surface defect inspection and classification are done manually, with low efficiency and reliability. In this paper, a deep learning method based on convolutional neural network feature extraction is used to realize the object detection and classification of metal shaft surface defects. Through image segmentation, the system methods setting of a Fast-R-CNN object detection framework and parameter optimization settings are implemented to realize the classification of 16,384 × 4096 large image little objects. The experiment proves that the method can be applied in practical production and can also be extended to other fields of large image micro-fine defects with a high light surface. In addition, this paper proposes a method to increase the proportion of positive samples by multiple settings of IOU values and discusses the limitations of the system for defect detection.

Purpose: For the climbing rod object with large diameter variation and the need of obstacle crossing, this paper aims to propose a new embracing-type climbing robot named as EVOC-I robot. Design/methodology/approach: The design philosophy and structural scheme are introduced. The kinematic analysis of embracing and telescoping mechanisms is carried out to provide the theoretical foundation for the effective climbing of the robot. Based on the prototype robot, three preliminary experiments are carried out to verify the effectiveness of the designed robot. Findings: The theoretical and experimental analyses have verified the reasonability and effectiveness of the proposed robot design. Research limitations/implications: As the preliminary study, the prototype still need a lot of improvement. The experimental verification is also limited. Future work will focus on improving the design and increasing the theoretical analysis, especially increasing experimental study and designing the next generation of the rod climbing robot. Practical implications: The designed climbing robot can be used for climbing the rod with variation diameter and flange obstacle, especially the lightening rod in the transformer substation. Originality/value: The paper designs a new climbing robot that integrates the ability of large variation diameter adaptation and obstacle crossing.

A methodology for determining the maximal regular-shaped dexterous workspace (MRsDW) of parallel manipulators (PMs) is proposed. The MRsDW is the maximal subset of the dexterous workspace with prescribed dexterity, and has a regular shape such as a circle or rectangle for the planar and a cuboid or cylinder for the spatial situation. It is one of the objects of optimal design of the PMs while the workspace size and shape are considered. The concept of the utilizable ratio of dexterous workspace (URDW) is introduced, which is a new measure for the rationality of design parameters of a PM. The methodology based on the stratified workspace boundary search technique can determine both the dexterous workspace and their MRsDW. After the problem of determination of the MRsDW is defined formally, a numerical method is proposed to solve this problem by three algorithms. Finally, the validity and efficiency of the proposed method are verified through application of the 2-DOF planar PM and the linear delta robot.

A methodology is proposed for determining the reachable and dexterous workspace of parallel manipulators. The determination of two types of workspace is utmost important in optimal design and comparing performances of the manipulators because each configuration or point in this region has specified kinematic dexterity by designer. The method is based on a stratified workspace boundary search technique, which is referred to as stratified workspace boundary determining methodology (SWBDM), and considers the various physical and contrived constraints. After the two similar problems of the determination of two types of workspace boundaries were defined, we propose a unified method to solve these problems. The validity and efficiency of the proposed method are verified with two kinds of representative parallel manipulators, since their relational results were studied in literatures. Finally, we conclude some advantages of the proposed methodology by comparing with other methods.