Visual inspection remains an essential tool for assessing structural damage. Damage detection is a challenging task for those specifying, designing, and deploying SHM systems. Often only traditional visual inspection processes are available to determine the type and extent of structural damage. For bridge structures in the UK, a regime of general (every two years) and principal (every six years) inspections is often followed. Such visual inspections are time-consuming and costly in terms of both labour and financial resources. Therefore, the possibility of completing more of the bridge visual inspection process offsite has many potential benefits for bridge owners and managers during the service life of the asset. Recent research conducted at the University of Bristol in collaboration with industrial partners has examined how to make the best use of metrics derived from visual inspection data when assessing bridge condition and planning maintenance activities. Recent research into which aspects of the current visual inspection regime in the UK could potentially be moved offsite has also been carried out. This paper summarises these research efforts and discusses how AI may be used as part of future enhancements to visual inspection data capture and analysis.

The visual inspection of bridges is a major undertaking for asset owners and operators. In the UK, visual inspections require inspectors to visit bridges on site and often at night and under unfavourable weather conditions. Therefore, it would be beneficial to move some of the visual inspection process off site. This paper studies whether the defect classification aspects of the inspection process could be conducted remotely using photographs. This study examines the defect ratings assigned by ten survey participants who were tasked with examining photographs from visual inspections of ten UK bridges. The survey results were compared with the results from the general inspections previously carried out for the bridges in question. From this data set, the differences in the ratings given and the extent to which defects are missed were examined. The results show that a higher number of defects were identified for a given bridge by the remote inspectors. Statistical analysis shows that aggregated defects rated by off-site inspectors tend to be more severe and of a higher priority rating compared with those from the on-site inspectors. The results also indicate that there is closer agreement between on-site and off-site inspectors for defects of a higher severity rating.

Visual inspection remains key for assessing the condition of bridges and hence assisting with planning and maintenance activities. There have been many efforts to improve or supplement visual inspection processes using new sensing technologies and data capture methods to usher in an era of 'smart bridges' or 'smart infrastructure'. One method to improve data capture is a 'remote inspection' where inspectors use digital photographs of a bridge to identify and grade structural defects to the standard of a 'General inspection' (GI). In this paper, survey data is presented to help formulate a preliminary assessment of the potential for engineers to implement this possible evolution of the visual inspection process. A potential Schema for remote visual inspections is developed and presented as a conceptual web application. The focus on the development of the Schema includes the need for ease of use by inspectors and integration of collected digital data into bridge management systems. The suggested platform is seen as a transitional method to aid in the long-term implementation of further automation of the inspection process. The system architecture is provided along with possible technologies that may support or enhance it, as well as a discussion of the potential barriers to implementation.