This post is also available in: heעברית (Hebrew)

A technology based on artificial intelligence and computer vision is helping monitor critical infrastructure, moving AI compute to the edge.

FirstEnergy has completed a pilot program demonstrating how computer vision can be deployed to analyze thousands of utility pole infrastructure images. The company teamed up with Noteworthy AI to install smart cameras in its utility truck fleet, with software powered by edge AI chips by Nvidia.

The platform provides geolocation for each pole in the widespread utility pole network before visually picking out the presence of components like insulators and current transformers. The computer vision software can then gauge whether it is physically damaged.

Nvidia says manual maintenance workers inspect a fraction of the 185 million utility poles in the U.S. in a single year. It would take an entire decade for them to inspect all of them. 

In a pilot test last summer, the technology collected more than 5,000 high-resolution images of its poles within 30 days, which expanded its database by more than fivefold.

Superior image quality is also anticipated to help avoid wasted visits by engineers to locations where the actual line conditions differ from initial expectations.

Since completing the initial test, a subsidiary of FirstEnergy focused on streetlights joined the program along with a unit that tracks vegetation growth around the company’s power infrastructure. The follow-up pilot covers more utility trucks over a larger area.

The smart camera module attaches to the truck with magnets or suction cups and links to a smaller unit inside the truck’s cab that processes the images, as reported by iotworldtoday.com.