Case Studies on AI-Assisted Sewer Condition Assessment of CCTV Data

CCTV condition assessment programs, which use robotic camera equipment deployed to inspect pipe segments one at a time, form the backbone of asset management and maintenance operations for sewer and water pipeline systems. Since 2002 NASSCO, the National Association of Sewer Service Companies, has offered the Pipeline Assessment Certification Program (PACP) training and certification courses for CCTV operators, asset managers, and engineers. PACP has quickly become the standard for sewer and storm condition assessments in North America, providing essential information decision-makers use to prioritize and optimize limited resources.
Unfortunately, the PACP is limited by human error and the wide variations in quality of PACP reports produced by different contractors, engineers, and operators in public agencies, not to mention the current workflow of human operators watching every foot and second of video in real-time contributes to approximately one quarter of the total cost to inspect and assess each foot of pipe.
An Artificial Intelligence-based (AI) tool to automatically label defects in sewer inspection videos using the existing PACP standards for defects and features in pipes has recently been developed. This development more than triples the speed of producing condition reports on wastewater, storm, and drinking water assets, while also reducing the total costs of inspection by approximately 50%.
The presentation will describe in detail the AI/Machine Learning processes involved in creating the PACP condition assessment reports using raw CCTV in-pipe data, as well as the major useful applications for the technology.

By Eric Sullivan