A few weeks ago I had the opportunity to talk with the co-founders of Buzz Solutions, a company that provides a platform to utilities for analyzing, storing, and managing the increasing load of high-resolution inspection images of transmission and distribution assets.
The CEO, Kaitlyn Albertoli, studied international relations, finance, and business at Stanford, a ran a non-profit around sustainable food with 60 people overseeing the needs of 300 people. Albertoli is from San Clemente, a beach town in southern California, and the San Onofre nuclear power plant was there. A big part of her upbringing were the controversial and freighted attempts to upgrade it and eventually shut it down. She became fascinated with energy and especially the renewables space.
The CTO, Vik Chaudhry, came from a diametrically opposed background, New Delhi in India. His Delhi College of Engineering undergrad was in civil and environmental engineering. In his final year he built a quadcopter from scratch with pollution monitoring sensors and flew it around hot spots in New Delhi. He found that it was often above 500 AQI, the equivalent of smoking 60 cigarettes a day. That air quality could be directly traced to a lack of electrification. He shifted to Stanford for a Masters focused on the energy sector, energy engineering, and application of AI and machine learning to demand response and energy efficiency, with a big project being drone-based wind farm siting assessments. Cisco scooped him up to run ML and AI for a few years as he and Kaitlyn got Buzz Solutions off the ground.
They met in a Stanford course in 2017, now called Venture Creation for the Real Economy. The first three weeks were setting up the go to market strategy, the next three weeks building your 5-year projection, and the last three weeks building your pitch deck. Stanford brings CMOs, CEOs, and VCs through every three weeks to give feedback.
Originally, they focused on wind farm siting, and rapidly were pointed to wind farm drone inspections. As they talked to energy companies, the narrative was always the same: have you seen what’s happening with more frequent inspections and the use of drones there? That led to them leveraging Stanford’s alumni lists to speak to 35 major power companies’ inspection teams. They found that they were capturing 10x the number of images as they were historically on a much more frequent basis and using drones a lot more. They were getting hundreds of thousands of high-res, high-zoom images with plans to expand to millions annually of their transmission and distribution infrastructure.
There is a lot more complexity, a lot more components, and a lot more observable issues in electrical transmission and distribution infrastructure than in wind farms. Severe faults could cause downed power lines, sparking and possibly forest fires. The concerns were leading to regulatory pressure for increased inspections and maintenance. And, of course, the infrastructure is aging.
But the process of assessing the data was manual, with trained linespeople and engineers looking at their computers instead of fixing problems. That was a machine learning opportunity. They started trying to assist utilities to identify hotspots and failure spots that could lead to significant outages. And that was before the 2017 wildfire season that damaged so many lines in California. They were there, they were ready and they were able.
The increases in inspections are driven by two major things. The infrastructure is aging, with the average age of components such as insulators being over 40 years old, and with limited asset tracking. A big part of what utilities are trying to figure out is what assets they have where, how old they are, and how fast they degrade. With renewables coming on the grid and climate impacts, components are going through increased stresses leading to increased degradation.
As renewables come onto the grid, modernization is a requirement, and identifying the components to be modernized is challenging. But you don’t tend to hear about the grid unless there’s a problem such as wildfire outages, rolling blackouts, or storm damage outages on the east coast. The grid has come into the spotlight in the past few years and serious attention is being paid to this aging and critical infrastructure in the US.
It isn’t possible to train 10x the people to deal with 10x the photos. Utilities were struggling with the problem. They were trying to hire engineers and linespeople to analyze the data, but the lag time of imaging to analysis was increasing. That meant an increase in risk, as degrading components saw more degradation in the meantime. Utilities became increasingly interested in solutions to manage and analyze the masses of data.
Buzz Solution’s market timing was excellent. High-resolution cameras on smaller, inexpensive drones allowed much more imagery, much more cheaply and more safely. A lot of UAV players have entered the market in the past 5-8 years, learning from their mistakes. Like DJI drones, they’ve all become smaller, more powerful, more intelligent, and sensors have shrunk radically. Fist-sized sensors are thumb-sized now. They are highly accessible platforms flyable by anybody.
And in 2016-2017, open source visual recognition machine learning toolkits became highly leverageable. Google released ImageNet and ResNet that rapidly became a standard backbone for image processing solutions. The algorithms being open source is key, but there has been great innovation in compute, with Cloud-based, cheap graphics processing units (GPUs) accessible in seconds with a few clicks. The result was a toolkit that was exploitable without a PhD in machine learning and a decade of use.
Historically, line inspections were done with helicopters or walking the lines with binoculars, and it was only a fourth or a third of the lines a year. There are also on-demand inspections when there are major storm risks, storm damage assessments, high winds, or risks of wildfires. Utilities identify hotspot areas prone to higher risk or where critical failures could occur, and those areas are inspected more frequently. There are both high-level flybys and more granular inspections. Obviously, more granular and more images requires more assessment.
For a single transmission tower, there can be 30 to 90 images per tower, usually in the 40-60 range. Even for distribution poles, the wooden or concrete ones bringing electricity to buildings, there are 4-12 images captured, typically in the 8-10 range. And there are a lot of towers. There are about 120,000 miles of high-voltage transmission lines in the US alone, with towers every fifth of a mile or so, suggesting about 600,000 towers. The distribution grid is vastly bigger, with about 5.5 million miles of wires and a lot more poles.
The data volumes are staggering, the need to inspect them is increasing significantly, and there aren’t enough people to do the work. Enter Buzz Solutions. The company saves 50% of image assessment time today, and is trending to 80% of effort savings.
The next era of inspections that’s emerging is neural net chips on drones, initially for automated flight around towers, identification of salient components and image capture, and eventually immediate fault detection. The Skydio drone, as an example, already uses an ML chip for its autonomous flying and imaging. With FAA regulations loosening and more companies getting approval for beyond-line-of-sight operations, there’s a fair amount of work in this space.
Buzz Solutions is well situated for this. They think of their product as an AI orchestration platform. It takes in multiple data streams and produces results. An extended use case can be a processing unit for an in-the-field drone. Various utilities and organizations are testing this now. The first step is recognizing assets with the drone following the power line itself, flying around the pole, and imaging them. The next generation will be heavier machine learning models with recognition of problems, and circling the poles to inspect components. If an insulator or conductor is observed to be damaged, the drone could circle it and take more images, and potentially send an immediate alert for prioritized fixing.
Buzz Solutions is fielding requests from its customers for this advanced functionality and will be deploying it when there is more leniency from the FAA on autonomy and more compute on the drones themselves. Vik’s love of building and flying drones will be satisfied again.
Another use case being explored is fixed-wing vertical takeoff drones which have longer ranges and higher resolution cameras so that they can capture higher res images at higher speeds. Quadcopters will still be needed for in depth granular inspections, but many third-party drone companies and utilities are looking at different form factors of UAVs for different use cases. This is one of many places where drones are displacing helicopters at vastly lower expense, part of the reduction in market for manned rotorcraft disrupting the industry.
And so ends the summary of the first half of my conversation with Vik and Kaitlyn, the co-founders of Buzz Solutions. Stay tuned for part two, dropping soon.
Don't want to miss a cleantech story? Sign up for daily news updates from CleanTechnica on email. Or follow us on Google News!
Have a tip for CleanTechnica, want to advertise, or want to suggest a guest for our CleanTech Talk podcast? Contact us here.