Electric power line patrol operation based on vision and laser SLAM fusion perception

Leiqing Ding,Jianjun Wang,Yikai Wu
DOI: https://doi.org/10.1109/auteee52864.2021.9668784
2021-11-19
Abstract:Transmission line inspection is one of the main responsibilities and tasks of the power grid operation and maintenance department. The current power grid high voltage transmission line inspection method mainly through maintenance personnel rely on ground transportation and low altitude manned helicopters, the use of handheld instruments or the naked eye to inspect facilities to deal with defects, its labor intensity, difficult working conditions, low labor efficiency, management inconvenience, difficult to adapt to the development of modern power grid and safe operation needs. The UHV and EHV grids are in dire need of safe, advanced and efficient power inspection methods. A further step forward is the use of large UAVs carrying high powered LiDARs to track lines by performing point cloud cuts, which are relatively costly for a single overhaul operation. This thesis mainly thinks about how to use the existing SLAM technology and vision algorithm combined with UAV to efficiently complete the power line inspection operation. The high-precision mapping of the area can realize the tracking and avoiding obstacles of the UAV to the wire. The main research idea of this paper is to analyze the current situation of the UAV line inspection work in the live operation and the real-time high-precision positioning and mapping operation, analyze the research status of laser SLAM and semantic SLAM, and further explain the use of Deeplabv3+ visual semantic segmentation network and dual sensor joint calibration.
What problem does this paper attempt to address?