电缆隧道环境下工业智能巡检机器人定位研究

doi:10.12068/j.issn.1005-3026.2025.20240212

摘要/Abstract

摘要：

电缆隧道封闭狭长，存在重复布设的电缆架和相似的场景纹理，属于退化场景.针对该场景，提出了1种基于点线特征融合的视觉惯导SLAM(simultaneous localization and mapping)算法.该算法通过长度抑制和短线拟合来改进高维线特征，使其能够更有效地描述结构化显著的隧道场景.此外，针对电缆隧道中特征相似导致的回环检测失败问题，引入具有高效识别和精确位姿估计的ArUco标记，限定回环发生区域，并利用最小化位姿变换筛选最佳回环帧，从而提升回环检测的准确度和定位精度.最后，在电缆隧道内采集数据集并进行实验验证.结果表明，相对于VINS–Mono(visual inertial system–Mono)，所提算法的绝对轨迹精度平均提升了69.73%，满足了电缆隧道巡检的应用需求.

关键词: 同步定位与地图构建, 点线特征融合, 回环检测, 工业机器人, 电缆隧道

Abstract:

The cable tunnel is closed and narrow， with repetitively laid cable racks and similar scene textures， which is a degraded scenario. To address this environment， a visual-inertial SLAM （simultaneous localization and mapping） algorithm based on point-line feature fusion is proposed. The algorithm improves the high-dimensional line features through length suppression and short line fitting to make it more effective in describing the structural features of tunnel scene. In addition， for the problem of loop closure detection failure due to feature similarity in cable tunnels， ArUco markers with efficient recognition and accurate pose estimation are introduced to limit the loop closure area， and the optimal loop closure frames are selected using the minimized pose transformation to improve detection accuracy and localization precision. Finally， dataset collection and experimental validation were conducted in actual cable tunnels. The results show that the absolute trajectory accuracy of the algorithm is improved by 69.73% on average relative to VINS–Mono（visual intertial system-Mono）， which meets the application requirements of cable tunnel inspection.

Key words: simultaneous localization and mapping, point-line feature fusion, loop closure detection, industrial robot, cable tunnel

中图分类号:

TP 242

王玉涛, 安俊炜, 秦长生, 郭伟帆. 电缆隧道环境下工业智能巡检机器人定位研究[J]. 东北大学学报（自然科学版）, 2025, 46(7): 49-58.

Yu-tao WANG, Jun-wei AN, Chang-sheng QIN, Wei-fan GUO. Research on Localization of Industrial Intelligent Inspection Robots in Cable Tunnel Environment[J]. Journal of Northeastern University(Natural Science), 2025, 46(7): 49-58.

图/表 16

图1 定位算法整体框架注：VIO(visual-inertial odometry);SFM(structure from markup).

Fig.1 Overall framework of positioning algorithm

图2 电缆隧道环境下ORB特征点提取效果

Fig.2 Effectiveness of ORB feature point extraction in cable tunnel environment

图3 基于RANSAC算法的误匹配滤除效果

Fig.3 Effectiveness of mismatch exclusion based on RANSAC algorithm

图4 ILSD算法流程图

Fig.4 Flowchart of ILSD algorithm

图5 线段误拟合（a）—误拟合前线特征；（b）—误拟合后线特征.

Fig.5 Wrong fitting of line segment

图6 改进后线段拟合结果

Fig.6 Improved fitting result of line segment

图7 基于LBD描述子的线特征匹配

Fig.7 Line feature matching based on LBD descriptor

表1 线特征追踪耗时对比

Table 1 Line feature tracking time consumption comparison

提取算法	线段总数/条	提取时耗/ms	描述时耗/ms	匹配时耗/ms	总时耗/ms
LSD算法	784	94.855	73.476	86.543	254.874
ILSD算法	150	36.238	28.172	4.751	69.159

图8 回环误检测结果

Fig.8 Wrong loop closure detection results

图9 ArUco标记图案

Fig.9 ArUco marker pattern

图10 ArUco标记多帧回环检测

Fig.10 ArUco marker multi-frame loop closure detection

图11 回环匹配帧对集合筛选策略

Fig.11 Filtering strategy of loop closure matching frame pair ensemble

图12 ArUco标记位姿测量模型

Fig.12 ArUco marker pose measurement model

图13 机器人运动坐标示意图

Fig.13 Schematic of robot kinematic coordinates

表2 直道和弯道数据集下评价指标对比 (m)

Table 2 Comparison of evaluation metrics under straight and curve datasets

评价指标	直道数据集			弯道数据集
评价指标	d	$d a v g$	$d m i d$	d	$d a v g$	$d m i d$
VINS-Mono/no loop	0.572	0.234	0.225	1.095	0.718	0.745
Ours-SLAM/no loop	0.353	0.143	0.140	0.862	0.454	0.493
Ours-SLAM /loop	0.222	0.138	0.132	0.238	0.130	0.131

表2 直道和弯道数据集下评价指标对比 (m)

Table 2 Comparison of evaluation metrics under straight and curve datasets

评价指标	直道数据集			弯道数据集
评价指标	d	$d a v g$	$d m i d$	d	$d a v g$	$d m i d$
VINS-Mono/no loop	0.572	0.234	0.225	1.095	0.718	0.745
Ours-SLAM/no loop	0.353	0.143	0.140	0.862	0.454	0.493
Ours-SLAM /loop	0.222	0.138	0.132	0.238	0.130	0.131

图14 轨迹结果比较（a）—直道数据集轨迹对比；（b）—弯道数据集轨迹对比.

Fig.14 Trajectory results comparison

参考文献 21

[1]	Mur-Artal R， Montiel J M M， Tardós J D. ORB-SLAM： a versatile and accurate monocular SLAM system［J］. IEEE Transactions on Robotics， 2015， 31（5）： 1147-1163.
[2]	Shan T X， Englot B. LeGO-LOAM： lightweight and ground-optimized lidar odometry and mapping on variable terrain［C］//2018 IEEE/RSJ International Conference on Intelligent Robots and Systems （IROS）. Madrid： IEEE， 2018： 4758-4765.
[3]	Mourikis A I， Roumeliotis S I. A multi-state constraint Kalman filter for vision-aided inertial navigation［C］//Proceedings 2007 IEEE International Conference on Robotics and Automation. Rome： IEEE， 2007： 3565-3572.
[4]	Qin T， Li P L， Shen S J. VINS-Mono： a robust and versatile monocular visual-inertial state estimator［J］. IEEE Transactions on Robotics， 2018， 34（4）： 1004-1020.
[5]	Dang T， Mascarich F， Khattak S， et al. Autonomous search for underground mine rescue using aerial robots［C］//2020 IEEE Aerospace Conference. Big Sky： IEEE， 2020： 1-8.
[6]	Azpurua H， Campos M F M， Macharet D G. Three-dimensional terrain aware autonomous exploration for subterranean and confined spaces［C］//2021 IEEE International Conference on Robotics and Automation （ICRA）. Xi’an： IEEE， 2021： 2443-2449.
[7]	Campos C， Elvira R， Rodríguez J J G， et al. ORB-SLAM3： an accurate open-source library for visual， visual-inertial， and multimap SLAM［J］. IEEE Transactions on Robotics， 2021， 37（6）： 1874-1890.
[8]	Cao S Z， Lu X Y， Shen S J. GVINS： tightly coupled GNSS-visual-inertial fusion for smooth and consistent state estimation［J］. IEEE Transactions on Robotics， 2022， 38（4）： 2004-2021.
[9]	Tang H L， Niu X J， Zhang T S， et al. LE-VINS： a robust solid-state-LiDAR-enhanced visual-inertial navigation system for low-speed robots［J］. IEEE Transactions on Instrumentation and Measurement， 2023， 72： 8502113.
[10]	Song B Y， Yuan X F， Ying Z M， et al. DGM-VINS： visual-inertial SLAM for complex dynamic environments with joint geometry feature extraction and multiple object tracking［J］. IEEE Transactions on Instrumentation and Measurement， 2023， 72： 8503711.
[11]	Li M L， Zhang H， Shen T A， et al. SM-VINS： a fast and decoupled monocular visual-inertial sensors SLAM system with stepwise marginalization［J］. IEEE Sensors Journal， 2024， 24（20）： 33240-33251.
[12]	Zhou H Z， Zou D P， Pei L， et al. StructSLAM： visual SLAM with building structure lines［J］. IEEE Transactions on Vehicular Technology， 2015， 64（4）： 1364-1375.
[13]	Forster C， Pizzoli M， Scaramuzza D. SVO： fast semi-direct monocular visual odometry［C］//2014 IEEE International Conference on Robotics and Automation（ICRA）. Hong Kong： IEEE， 2014： 15-22.
[14]	Gomez-Ojeda R， Briales J， Gonzalez-Jimenez J. PL-SVO： semi-direct monocular visual odometry by combining points and line segments［C］//2016 IEEE/RSJ International Conference on Intelligent Robots and Systems （IROS）. Daejeon： IEEE， 2016： 4211-4216.
[15]	Fu Q， Wang J， Yu H， et al. PL-VINS： real-time monocular visual-inertial SLAM with point and line features［J］. arXiv preprint arXiv， 2020： 2009.07462.
[16]	Lim H， Kim Y， Jung K， et al. Avoiding degeneracy for monocular visual SLAM with point and line features［C］//2021 IEEE International Conference on Robotics and Automation （ICRA）. Xi’an： IEEE， 2021： 11675-11681.
[17]	Zhu Y Q， Jin R， Lou T S， et al. PLD-VINS： RGBD visual-inertial SLAM with point and line features［J］. Aerospace Science and Technology， 2021， 119： 107185.
[18]	Lee J， Park S Y. PLF-VINS： real-time monocular visual-inertial SLAM with point-line fusion and parallel-line fusion［J］. IEEE Robotics and Automation Letters， 2021， 6（4）： 7033-7040.
[19]	Rublee E， Rabaud V， Konolige K， et al. ORB： an efficient alternative to SIFT or SURF ［C］//2011 IEEE International Conference on Computer Vision. Barcelona： IEEE， 2011： 2564-2571.
[20]	Von Gioi R G， Jakubowicz J， Morel J M， et al. LSD： a line segment detector［J］. Image Processing on Line， 2012， 2（4）： 35-55.
[21]	Hamid N， Khan N. LSM： perceptually accurate line segment merging［J］. Journal of Electronic Imaging， 2016， 25（6）： 061620.