UAV Inspection Service Reveals the Power of Artificial Intelligence and Machine-Learning

May 14, 2019 | Author: Brad Westphal

Honeywell’s unique expertise in advanced data analytics is helping utilities improve safety and efficiency by providing actionable insights that help them maintain the powerlines and power poles they use to deliver safe, reliable electric service to their customers.

Our UAV inspection and data analytics service uses unmanned aerial vehicles (UAVs) to perform inspections usually made by helicopters, small aircraft or walking crews. With UAVs – often called drones – asset inspections can be accomplished in a fraction of the time by fewer people. As a result, workers, pilots and aircraft are exposed to fewer safety hazards. The inspections are more thorough and complete, since drones fly lower and slower, capturing detailed images used to catalog pole damage and vegetation encroachment.

The use of drones represents a significant advancement in the inspection process across several industries including utilities, oil and gas and renewables. But the real breakthrough comes from using data analytics software, made up of artificial intelligence and machine-learning algorithms, to analyze the images the drones capture and convert them into data that is timely, accurate and – most importantly – actionable.


Rather than using generic off-the-shelf analytics packages, Honeywell’s customer-centric approach enabled us to quickly deploy tailored and scalable solutions that meet our customers’ specific needs. Honeywell is also utilizing innovative ways to shorten the development time while maintaining the quality of our offerings. For instance, we’ve created an advanced photo-realistic simulation framework that lets us synthesize representative data to augment real-world data thus reducing the need for expensive field-data collection. We also use this simulation framework to quickly validate and iteratively improve our analytics. Of course, more representative data always improves accuracy.

We also require fewer images than alternative solutions – less than a couple of hundred images compared to the thousands needed by conventional machine-learning methods. This allows us to quickly deploy models and get valuable feedback from customers, sooner rather than later.

As always, cybersecurity is a top priority for Honeywell. We’re constantly looking at ways to protect and secure our customers’ data by leveraging cloud providers’ capabilities and adding our own authentication service, which is General Data Protection Regulation compliant.


The result of all this is a more robust solution that customers can trust, based on our century-long history of achievement in aerospace, our extraordinary expertise in data analytics, our subject matter expertise and large customer base.

Ozarks Electric Cooperative, an electric distribution co-op with more than 70,000 residential, commercial and industrial members in northwest Arkansas and northeast Oklahoma, uses our data analytics service and they’ve been very pleased with the results.

We’re currently working with Ozarks Electric to conduct a second biannual inspection using drones to examine 150 miles of powerlines and more than 2,600 poles, many of which carry electric power through rough and rugged terrain that is challenging and time-consuming to negotiate on foot. Historically, it’s taken a crew of seven people 2-3 weeks to complete the inspection. Inspectors frequently encounter venomous snakes, wild pigs and other dangerous critters along their inspection route.

In October 2018, Honeywell UAV Commercial Service captured about 168,000 images for the Ozark Electric team in just five days. Impressive. But, it would be impossible for the team to manually go through that many images to look for pole damage, vegetation encroachment and other potential problems.

So we applied our proprietary data analytics software to reduce the number of images and identify about 6,000 images showing poles that appeared to be cracked or rotted and poles or lines with vegetation issues. Six thousands is a much more manageable number, but still too many for human beings to efficiently analyze. We knew we could do better.


We went back to the Ozarks Electric team to tighten the inspection parameters. Then we trained the inspection-system software using new metrics to make the output more accurate and actionable. We tweaked the algorithms to recognize the difference between a pole showing signs of rot and those just discolored by bird droppings and to look exclusively for cracks that ran against the grain of the wood.

With these kinds of instructions we were able to hand Ozarks Electric a report that identified just 67 poles needing maintenance and 25 cases of vegetation encroachment. The software also ranked the issues, determining which ones were “urgent” and needed immediate attention and which were “important,” so the utility could set priorities and address the most critical problems first. Other situations were identified as needing to be monitored.

Ozarks Electric has been a great partner as we continue to improve the capabilities of our UAV inspection service and deliver even greater value to our customers. Their input has been extremely valuable in developing a system that provides the ultimate inspection experience for utilities and customers in other industries.

Let us know how we can help you keep an eye on your critical assets.


Brad Westphal

Senior Director of Commercial UAV Solutions at Honeywell Aerospace


  • Oscar Valdez

    I’m interested in Oil&Gas pipelines inspections , to get parameters and risk analysis.

  • Mark

    Brad, curious to see if your solutions are using IR and whose the IR core provider. My larger curiosity is on the oil and gas services side and if your providing gas detection via UAS. Can we have a conversation on the aforementioned? Thank you in advance for your time. Mark Fishburn NSM Gas Detection / Digital Oil Field FLIR Systems (970) 646-3289