Traditional sewer inspection methods, such as internal CCTV, can be time-consuming and can overlook defects due to inaccurate identification and subjectivity in how people code. When assessing the future degradation of a sewer line or determining which asset to prioritize for rehabilitation, the difference and accuracy of coding is critical.
The goal of integrating artificial intelligence (AI) with sewer inspection is to supplement workers in the field, not to replace them. AI takes on the more common defects, allowing field workers to focus on work at hand (access, MOT, cleaning) and codes that are more difficult and less frequent.
As with any project, the more accurate the data that goes in, the better quality of data that comes out. It is necessary to capture clear and unobstructed video, whether a field technician or an AI-based platform performs the assessment. When provided clear video, B&N’s AI has an accuracy rating of approximately 90%.
Our computer vision AI system incorporates machine learning algorithms that capture and recreate the workflow and thought process a person works through in coding and performing the quality assurance/quality control (QA/QC) of sewer inspections. With this technology, we have substantially reduced the time required to review sewer inspection data, increased the number and accuracy of defects identified and coded, and supplemented the human element that is prone to bias. Removing the burden of coding all defects from the contractor allows them to inspect more footage in a day and reduce the cost-per-foot for the owner. Utility providers can quickly move through existing video to provide a database that shows the condition of their storm and sanitary infrastructure, allowing them to make data-driven decisions that are transparent and repeatable.
This presentation will demonstrate the benefits of using AI as a low-cost way to evaluate systems and better maintain assets to prioritize rehabilitation and coordinate with other capital improvement projects.