ASDER Project

Traditional sewer inspection process is inefficient, causing many sewer inspection companies to struggle with backlog (Municipal Sewer&Water Magazine,2022). It involves sending cabled robots with CCTV-cameras into the sewer, controlled by a sewer inspector remotely from a van with access to the live videos. The controller moves the robot slowly, pans the camera circumferentially to check for defects at every metre, and documents defects using Wincan software. This process takes 10-days for one-kilometre sewer length, plus a day to produce inspection report from the video/Wincan document. Sewer companies’ thus commonly struggle with backlog. A southern city currently has backlog of 3000-miles of sewer (Municipal Sewer&Water Magazine,2022).

Wincan software (existing alternative) only lets an inspector capture (photo) and annotate a defect with the right name and grade directly on the live video from the robot, making subsequent report preparation easy. ASDER will develop Artificial-Intelligence and BigData models that will automatically identify defects from sewer inspection videos and produce reports with defects photos, each correctly labelled with its name and grade, in 5-minutes. Report will contain pictures of identified defects, defects’ location, and minutage point where specific defects appear in the inspection video. Inspectors can use this information to audit the system by checking if the said defects exist in the claimed location and correctness of name and grade. ASDER will also use Federate Learning to preserve users’ data privacy. These will provide assurances to users.