Leveraging AI, Video, and GPS for Distribution Inspections

To address these challenges, we present an AI‑assisted, video‑ and GPS‑based drive-by inspection method. Unlike aerial approaches, our method uses cameras mounted on standard vehicles to continuously record videos along distribution routes. This setup provides several advantages: it requires only a single driver, eliminates the need for defect knowledge during data collection, improves safety, accelerates inspection, and reduces operational costs.

Specifically, each video stream is processed frame by frame by a pole detection model, and detected poles are assigned unique identifiers through a tracking algorithm. Embedded GPS metadata are synchronized with video frames to capture vehicle geolocation at a fixed frequency. For each pole detected within a predefined distance, the system estimates the pole’s geographic coordinates from the vehicle’s GPS data and applies specialized AI models to detect specific defects. To streamline review and documentation, we develop a human-in-the-loop review application that enables rapid validation and automated report generation. Our drive-by inspection method has been successfully tested in field pilots across multiple feeders, demonstrating a practical and scalable path toward field-deployable distribution inspections.