基于联动控制与AI视觉识别技术的地下管廊智能巡检机器人研究  

Research on Intelligent Inspection Robot Technology for Underground Pipe Corridors Based on Linkage Control and AI Visual Recognition

作　　者：田洪伟 撒兴才 李钰 唐戬 TIAN Hongwei;SA Xingcai;LI Yu

机构地区：[1]北京普龙科技有限公司,河北雄安新区071800

出　　处：《科技创新与应用》2025年第10期10-13,共4页Technology Innovation and Application

基　　金：2023年度河北省高水平人才团队建设专项产业创新创业团队项目(235A1802D)。

摘　　要：随着智慧城市建设的推进,地下综合管廊的运维需求不断增加。传统的人工巡检不仅效率低下,还存在安全隐患。为此,该文提出一种集成联动控制与人工智能视觉识别技术的智能巡检机器人系统。该系统通过将YOLOv5算法与深度强化学习算法相结合,实现对火焰、烟雾、积水等目标的高效识别和自动巡检路径优化。联动控制技术优化机器人在复杂环境中的运动和定位,使其能够精准适应地下管廊的复杂条件;同时,YOLOv5的多层特征融合机制提升目标检测的准确性和响应速度。实验结果表明,YOLOv5在目标检测中的准确率达到92.31%,展示该系统在提升巡检效率和安全性方面的显著优势。With the advancement of smart city construction,the maintenance and operation demands of underground duct banks have become increasingly complex.Traditional manual inspection methods face issues of inefficiency and safety hazards,leading to the emergence of intelligent inspection robots.This study explores an intelligent inspection robot based on linkage control and AI visual recognition technology to enhance the automation level of underground duct bank maintenance.By combining linkage control technology with the multi-task convolutional neural network(CNN)algorithm YOLOv5,the research designs and implements an intelligent inspection system.YOLOv5 demonstrates excellent accuracy and response speed in real-time identification of targets such as flames,smoke,and water accumulation.Experimental results show that the accuracy rate of YOLOv5 in target detection reaches 92.31%,demonstrating the significant advantages of the system in improving inspection efficiency and security.

关 键 词：联动控制 AI视觉识别 YOLOv5 目标检测 地下综合管廊 

分 类 号：TU990.3[建筑科学—市政工程]