This thesis focuses on the intelligent operation and maintenance requirements of urban underground utility tunnels

and designs and implements a set of intelligent detection and operation and maintenance system for urban underground utility tunnels based on a multimodal large model. To address the problems of low efficiency

high cost

and delayed warning in traditional manual inspections

this system uses an intelligent robot dog as the carrier

integrates multimodal sensing and 5G technology to achieve dynamic monitoring

uses the large model for intelligent diagnosis and warning

and realizes closed-loop management of operation and maintenance through digital twin technology

aiming to provide a lightweight and efficient solution.