Aerial T&D lines inspection with AI

Project Overview

Concept X took up an AI aerial inspection of distribution lines for a leading US electric utility. The objective was to demonstrate the power of AI in detecting a range of simple and complex defects, and recommend the optimal inspection protocols (flight path, payloads) for data acquisition. Three different distribution lines captured with multiple camera models (images + videos) were analysed and fed into the platform to identify, annotate and produce results.

Our Solution

We deployed our solution to automatically process visual data from multiple camera systems. Our Lambda Scan application provided recommendations on data usability for ML , along with optimal future inspection protocol. Through our Lambda Inspect software we were able to accurately identify a variety of defects, including:

1. High visibility damages such as woodpecker holes, cracks, pole cap issues, etc.
2. Very nuanced defects, such as missing or loose bolts and backed out cotter keys
3. Rare defects without enough training examples by using Gen AI methods

All results provided through interactive dashboard, along with ML diagnostics report.

Project Outcome

Concept X showcased the use of AI in both automation and processing.

Key benefits for the client:

• Opportunity identified to reduce flight times by 40% and decrease inspection spend
• 95%+ defect capture rate via AI, including on hard cases (loose bolts, missing keys)
• 20% additional, potential defects detected which were missed in original inspection