Energy
Exelon® Leads Utility Innovation With Deloitte in Autonomous Drones for Grid Asset Inspection
Deloitte, Exelon
Objective
As one of the largest energy companies in the United States serving over 10 million customers, Exelon was looking to deliver faster, safer, and more accurate grid asset inspections using drones and vision AI. In collaboration with Deloitte and NVIDIA, Exelon developed and deployed OptoAI—an autonomous drone solution built on the NVIDIA Jetson™ edge AI platform and NVIDIA Omniverse™. OptoAI turns time-intensive asset inspections into rapid-response solutions for Exelon and its customers. This work was led by Baltimore Gas & Electric (BGE®), Exelon’s Maryland-based utility, and supported by the BGE Innovation Council. Public Utilities Fortnightly announced that this innovation was honored with the Edison Pioneers Top Innovator Award for Collaboration in Innovation.
Customer
Exelon
Partner
Deloitte
Use Case
Computer Vision / Video Analytics
Products
NVIDIA Jetson
NVIDIA Omniverse
Key Takeaways
Elevating Grid Asset Insights in the Field
Every day, field workers at utilities ensure the safety and reliability of grid infrastructure by “walking the line” to identify defects and resolve safety concerns that can affect everything from vast forests to massive population centers. A critical function for power providers and grid system operators, asset inspections are as necessary as they are arduous and labor-intensive. It can be a high-impact, high-risk endeavor. Field workers may climb hundreds of distribution poles to inspect and resolve issues or defects. And while the adoption of drones has helped improve the scale and scope of their inspections, using AI for autonomous inspections can bring unprecedented advantages to field operations across transmission and distribution.
Designed to achieve more accurate inspections with faster response times and improved reliability, Exelon’s AI-driven autonomous inspection initiative is designed to put more power in the hands of operators and supervisors. The solution enables grid operators to plan and execute missions more efficiently, identify and inspect assets more rapidly, and immediately deploy work crews to fix the problematic issues it detects. The initiative also aims to set a new standard for safety and performance in inspections.
Deloitte, Exelon
Putting Autonomous Drones and AI to the Test
Traditional drones and sensors require a pilot to fly them, making sure to gather images on a shot list and then manually pass these images off to an inspector for analysis. The time from inspection to action plan can take weeks. By using OptoAI, an AI platform from Deloitte powered by NVIDIA for real-time asset inspection and autonomous drone flights, Exelon’s drones can operate independently to execute missions autonomously. Each drone can recognize assets, detect defects, and make informed recommendations to field crews on the repairs, such as damaged crossarms, insulators, lightning arrestors, transformers, or wires. As a result, Exelon can prioritize and fix issues immediately from reports that would have otherwise taken weeks to compile.
“At Exelon, we’re proud to support BGE’s leadership in pioneering this groundbreaking work. By combining autonomous drone technology with AI from Deloitte and NVIDIA, we’re transforming how we inspect and maintain our grid—making it faster, more accurate, and ultimately safer for our field teams and more reliable for our customers.”
Ankush Agarwal
Director, Innovation & Technology, Exelon
Deloitte, Exelon
Exelon’s uncrewed aerial systems manager wearing an augmented reality headset for virtual fly-through using OptoAI to inspect power lines, distribution poles, and other grid infrastructure.
How Does Exelon’s OptoAI Solution Work?
Since 2022, Deloitte, Exelon, and NVIDIA have been working together to deploy AI for field operations across Exelon’s service territories. This includes a three-step approach to implement such autonomous grid asset inspections, from drone training to real-time analysis and decision-making:
Step 1: Training AI Models
Advanced vision AI algorithms were trained on NVIDIA’s accelerated computing platform. Because there wasn’t enough grid-edge data to effectively train the algorithms, supplemental synthetic datasets were generated using NVIDIA Omniverse, a platform to develop generative physical AI-powered applications for industrial digitalization. The synthetic data was used to train the vision AI models for the drones using a wide range of inspection scenarios.
Learn more about this award-winning work in “Exelon Uses Synthetic Data Generation of Grid Infrastructure to Automate Drone Inspection.”
Step 2: Deploying AI on the Grid
The drones integrate an on-board module from NVIDIA Jetson, a platform for robotics and embedded edge AI applications. The Jetson module runs the vision AI models that allow the drones to autonomously locate and inspect assets (i.e., poles) for defects (i.e., broken crossarms or leaning poles).
Geographic information systems (GIS) asset location technology enables these drones to autonomously locate and inspect assets, as well as uncover missing asset locations, without any preprogramming. It can also improve asset integrity by auto-correcting GIS locations when mistakes are found.
The visual data that the drone collects as it flies over grid assets is processed at the edge rather than the cloud. This solution gives field workers a detailed inspection report on the spot via an interactive user interface, providing real-time AI insights without any delays.
Step 3: Designing Interfaces for Actionable Results
Field workers receive a detailed, color-coded inspection report via an interactive user interface on their tablet, while supervisors access real-time, actionable insights from their computer or extended reality (XR) headset. The findings are prioritized for grid operators to quickly deploy field crews to hot spots for immediate repairs. They can pinpoint the location of the grid asset and the area of concern at a glance.
Deloitte, Exelon
Enhancing Operational Efficiency With Autonomous Grid Asset Inspection
OptoAI underscores the use of AI to enhance field operations, paving the way for operational optimization and setting a new standard for other utilities and grid operators to adopt. This solution is set to decrease operational costs, expedite emergency responses, keep power on for customers, and raise operational efficiency, all while enhancing safety.
Potential benefits include:
- Faster Mission Planning: Reduces preflight planning time and streamlines processes, compressing drone operation time from up to an hour to 30 seconds or less—an over 100x speedup.
- Increased Speed and Accuracy: Speeds identification and analysis of components and defects using AI models with less room for human error.
- Real-Time Inspection Results: Streamlines the identification and sharing of critical defects with operators in real time via web, tablet, or XR. Filters thousands of daily photos to a select few critical images, saving operators weeks of sorting time.
- GIS Corrections: Facilitates data accuracy with automatic updates to GIS systems.
- Improved Safety: Reduces manual inspections, minimizing risk to human inspectors.
- Improved Customer Service: Protects customers from power outages with better, proactive maintenance and quick identification of defects.
Deloitte, Exelon, and NVIDIA are bringing AI to the grid-edge to enable more efficient field operations in the power and utilities industry. The integration of autonomous drones and AI in utility field operations helps field workers to cover more territory faster and with less risk, while utilities can achieve new levels of efficiency, accuracy, and safety.
Explore power and utility companies collaborating with NVIDIA to enhance grid resiliency and reliability with software-defined infrastructure.