基于EKF的变电站巡检机器人VIO/GPS融合定位算法  

VIO/GPS Fusion Positioning Algorithm of Substation Inspection Robot Based on EKF

作　　者：张贵峰 张忠轩 周朝荣 于智龙[4] ZHANG Guifeng;ZHANG Zhongxuan;ZHOU Chaorong;YU Zhilong(China Southern Power Grid Research Institute Co.,Ltd.,Guangzhou 510000,China;School of Instrument Science and Engineering,Harbin Institute of Technology,Harbin 150000,China;Wenshan Power Supply Bureau of Yunnan Power Grid Co.,Ltd.,Wenshan 663000,China;School of Automation,Harbin University of Science and Technology,Harbin 150080,China)

机构地区：[1]南方电网科学研究院有限责任公司,广东广州510000 [2]哈尔滨工业大学仪器科学与工程学院,哈尔滨150000 [3]云南电网有限责任公司文山供电局,云南文山663000 [4]哈尔滨理工大学自动化学院,哈尔滨150080

出　　处：《哈尔滨理工大学学报》2024年第5期48-55,共8页Journal of Harbin University of Science and Technology

基　　金：国家自然科学基金(61673128,61573117)。

摘　　要：随着电力事业的发展,变电站智能巡检系统被广泛应用以保障变电站的安全运行与供电质量。针对现有巡检机器人导航技术中定位精度不高的问题,提出了一种基于扩展卡尔曼滤波的视觉惯性里程计与GPS融合算法。该算法通过视觉惯性估计传播滤波器状态,并利用GPS测量进行更新,有效消除位置和航向累积漂移,同时对计算资源需求较小。理论和实验验证表明,该算法能够在室外变电站环境下实现全局一致且准确的六自由度姿态估计。With the development of the electric power industry,intelligent inspection systems for substations have been widely applied to ensure the safe operation of substations and maintain power supply quality.To address the issue of low localization accuracy in existing substation inspection robot navigation technologies,this paper proposes a fusion algorithm based on visual-inertial odometry and GPS using Extended Kalman Filtering(EKF).The algorithm propagates the filter state through visual-inertial estimation and updates it using GPS measurements,effectively eliminating cumulative drift in position and heading,while requiring minimal computational resources.Theoretical and experimental results demonstrate that this algorithm can achieve globally consistent and accurate six-degree-of-freedom pose estimation in outdoor substation environments.

关 键 词：视觉惯性里程计 全球定位系统 扩展卡尔曼滤波 变电站巡检 机器人导航 

分 类 号：TM63[电气工程—电力系统及自动化] TP242[自动化与计算机技术—检测技术与自动化装置]