Research on the Application of Binocular Vision-Based Pose Estimation Technology in Live-Line Working. Issue 1 (December 2020)

• Record Type: Journal Article
• Authors: Wang, Jianmin; Xu, Guangnian; Wu, Baijun; Dai, Jiangjun
• Is Part Of: IOP conference series. Volume 619:Issue 1(2020)
• Publication Date: 2020-12
• Subjects: Earth sciences -- Periodicals; Environmental sciences -- Congresses; Environmental sciences -- Periodicals; 550.5
• DOI: 10.1088/1755-1315/619/1/012009
• Languages: English
• ISSNs: 1755-1307

Abstract

Abstract: To engage robots in live-line working and reduce the cost of electric power companies, this study proposes a binocular vision-based pose estimation method that can automatically identify the target's pose, guide the robotic arms to grasp the tools and fulfill live-line working tasks. First, this method matches and calculates the deep information of images via the binocular camera, and converts the images into the cloud of points; then, the advanced image segmentation technology is used to recognize the tools, and 3D modelling is conducted for the objects to accurately match the target image and the real target; the iterative closest point (ICP) method is used to solve the spatial pose of the target corresponding to the robot and thus guide the robots in tasks. Meanwhile, experiments are conducted to simulate the actual work process, and the result shows that the proposed method can accurately estimate the pose of targets and has good robustness and application value.