AbstractTo ensure the safety of a concrete structure, periodic and accurate inspection for damage is necessary. However, there is a high risk of inaccurate diagnosis of the structure condition because the manual inspection method is dependent on operator experience and subjective judgment. Additionally, in future society, it may become increasingly difficult to maintain high‐quality infrastructure, as the number of inspection personnel will also decrease with a decrease in population. For this purpose, the demand for accurate and automated inspection technology is increasing. To meet these issues, in this study, an objective and automated inspection method is proposed by developing an inspection robot and management system. The inspection robot comprises an artificial intelligence sensor, a manipulator, and a ground vehicle; it can autonomously drive inside a tunnel and inspect the damage on concrete surfaces using stereo vision. The management system remotely controls the robot through wireless communication and monitors the inspection status in real time. In addition, damage is automatically restored in three dimensions based on the images, and the physical size of the damage is analyzed with an average relative error of 0.39%. This method is expected to be used as a core technology for unmanned and automated tunnel maintenance in the future.