Exelon has deployed OptoAI, an autonomous drone inspection system for power lines, distribution poles and transmission towers developed with Deloitte and NVIDIA, according to an NVIDIA customer story published in 2026. The work was led by Baltimore Gas and Electric, Exelon's Maryland utility, and NVIDIA reports that drone operation time per inspection fell from up to one hour to as little as 30 seconds.
"By combining autonomous drone technology with AI from Deloitte and NVIDIA, we're transforming how we inspect and maintain our grid," said Ankush Agarwal, director of innovation and technology at Exelon.
Conventional drone inspections need a pilot to fly a shot list and then hand images to an inspector, so the time from inspection to repair plan could take weeks, and field workers still climb hundreds of poles.
Deloitte, Exelon and NVIDIA have worked on the system since 2022 in three steps. Vision AI models were trained on NVIDIA accelerated computing, and because there was not enough real grid-edge imagery, synthetic training data covering many inspection scenarios was generated with NVIDIA Omniverse. On the drones, an onboard NVIDIA Jetson module runs the models at the edge rather than in the cloud, so each drone can recognize assets, detect defects such as damaged crossarms, insulators, lightning arrestors, transformers and wires, and recommend repairs. GIS asset-location technology lets drones find assets without preprogramming, discover missing asset records and correct wrong GIS locations. Field crews get a color-coded, prioritized report on a tablet, and supervisors see results on a computer or extended-reality headset, so thousands of daily photos are filtered down to a few critical images.
NVIDIA presents faster mission planning, fewer manual inspections and lower operating costs as benefits of the approach. The project received the Edison Pioneers Top Innovator Award for Collaboration in Innovation.
Details
- City
- Baltimore
- Organization
- Exelon
- Continent
- North America
- Country