A substation switch room operating robot end positioning method based on deep learning

By combining deep learning and computer vision methods, the image matching accuracy and point cloud processing of the substation switch room operation robot were improved, solving the problem of the robot's inability to accurately locate itself and achieving high-precision target recognition and safe operation.

CN115311433BActive Publication Date: 2026-05-26SHANGHAI WUJIN FIRE-FIGHTING SAFETY EQUIP CO LTD
View PDF 2 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
SHANGHAI WUJIN FIRE-FIGHTING SAFETY EQUIP CO LTD
Filing Date
2022-08-09
Publication Date
2026-05-26

AI Technical Summary

Technical Problem

Existing substation switch room operation robots cannot accurately and clearly see the equipment status, resulting in poor positioning accuracy or inability to locate, especially in complex backgrounds where a single algorithm cannot meet the requirements.

Method used

By combining deep learning models and computer vision, a matching algorithm combining SuperPoint and ASIFT is used to improve image matching accuracy, and a point cloud depth map leveling algorithm is used to eliminate the posture deviation of the robotic arm end effector, so as to achieve parallel positioning of the end effector and the target being operated.

Benefits of technology

It improves the accuracy of end-point positioning, enhances the safety factor of operation, and ensures the power supply stability and equipment safety of power grid companies.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN115311433B_ABST
    Figure CN115311433B_ABST
Patent Text Reader

Abstract

This invention relates to the field of end-effector positioning technology for substation switchgear operating robots, specifically a deep learning-based end-effector positioning method for substation switchgear operating robots. The method includes an end-effector recognition algorithm, input task parameters, vehicle position adjustment, end-effector posture adjustment, and target positioning. Compared with existing technologies, this invention improves the accuracy of image matching algorithms by using a SuperPoint+ASIFT combined matching algorithm, eliminates interference from deviations between the target plane and the robot arm's end-effector posture using a point cloud depth map leveling algorithm, and extracts the position and angle of the target on the image to be recognized to meet positioning accuracy requirements. This improves the accuracy of end-effector positioning, enhances the safety operation factor, ensures the stability of power supply from power grid companies to society, and guarantees the safety of the power grid and its equipment.
Need to check novelty before this filing date? Find Prior Art