UU rolls out AI sewer inspectors

United Utilities has begun inspecting wastewater pipe networks using artificial intelligence to review footage from CCTV within pipes.

The company said the semi-automated reviews speed up fault detection to directs teams to where repairs are required. Survey processing time has been reduced from 10 days to two for routine maintenance inspections.

CCTV has been used since the 1950s, but has required manual reviewing of thousands of kilometres of network for pipe damage or obstructions that could lead to sewer flooding.

A partnership with Vapar, an Australian AI company, was formed through UU’s Innovation Lab to improve the accuracy of the AI.

Katy Bevan, programme delivery manager for UU’s wastewater network said: “Vapar has truly transformed the way that we work, it provides us with a consistent method of defect coding that equates to a consistent decision-making process on investment.

“It has also massively reduced timescales of both the time needed to review footage and the turnaround time from footage being recorded on site to decisions about cleaning and remedial activities being made. In addition we are also benefiting from lower costs, and the generation of performance management tools to support our targets.”

The complex nature of pipeline inspections requires a “collaborative intelligence approach” with AI analysis still needing human review.

The technology is now being rolled out across UU’s network for use in all inspections and integrated with the company’s dynamic network monitoring platform, which contains feeds from thousands of monitors and datasets to proactively manage networks.

Vapar took part in the Innovation Lab in 2019 to develop its system and improve the accuracy of the AI as part of UU’s collaborative workshops designed to help innovative technologies get adopted faster.