Buzz Solutions: Transforming Asset Inspections with AI
Electricity and the Power Grid
Available at a flick of a switch, we often take electricity for granted, forgetting that it’s the lifeblood of our society and the economy. Electricity has always fascinated humankind. While ancient civilizations had not harnessed the power of electricity, the imposing power of lightning captivated their citizens, who often attributed it to divine anger. Ancient Greeks credited Zeus, while Vedic tradition associated lightning with Indra. And even though it would take millennia before the first public electricity supply, there’s no denying its power across all of recorded history.
Americans have their own lightning mythology: Benjamin Franklin’s famous kite experiment in 1752. As the legend goes, one stormy night, Ben Franklin strung a metal key to a metal kite wire and captured lightning as it struck. From here, the metal lightning rod was born and opened the floodgates of discovery. In the 100 years following the experiment, groundbreaking inventions such as the first battery and the telegraph were invented.
Then, in 1879, there was — the first — lightbulb moment, as Thomas Edison successfully designed the first working light bulb. Shortly after, Edison established the first central power station in New York City, and by 1925, half of all homes in the United States were illuminated by Edison’s groundbreaking innovation.
Electricity starts to play its role in our lives from the moment we’re born — silent and invisible but omnipresent. Were we to somehow transport one of those ancient Greeks to modern-day America, electricity and all it does would be nothing short of miraculous in their eyes.
It is the single ingredient that touches every aspect of our lives. Were it not for electricity you would not be reading this article; you would not be able to access the Internet, connect with people, or voice opinions on social media. There’s a very good chance that the majority of everyday products around you would not exist, and if they did, they wouldn’t be nearly as refined.
We’re so accustomed to the convenience provided by electricity, that we rarely stop to think how it works. Electricity is sent across long distances using high-voltage transmission lines, and local facilities known as substations convert that high-voltage power to a lower voltage and distribute it to nearby homes and businesses. Your job, your commute to and from work, your telecommute, time with your family, calls home, baseball, football and basketball games, heating, cooling, food production, healthcare, transportation, manufacturing, national security, the economy: all enabled by electricity.
Legacy Infrastructure and Inspection Methods
The U.S. electric grid dates back to 1882, the year that Thomas Edison unveiled the country’s first power plant at the Pearl Street Station in lower Manhattan. During the last 130 years since the first power line was installed in the US, the power grid in the country has grown explosively to meet the demand of the largest economy in the world. While the grid has greatly expanded, for decades its basic structure has remained much the same. The power grid in the United States depends on old and aging infrastructure; the average power pole for a distribution line is more than 40 years old. This introduces many points and types of failures, and any single fault can be disastrous.
Today, global utilities spend billions of dollars per year on asset inspection programs to reduce outages, increase compliance, improve safety, and reduce the probability of catastrophic events such as wildfires or major regional outages, which we’ve seen increasingly in recent years. Optimal asset functioning is always the goal. Yet, legacy survey models are very expensive, time consuming and can have unacceptable liability risk, which is why the Unmanned Aircraft Systems Rulemaking Committee is mandating that data collection moving forward must utilize drones.
Legacy inspections have been done manually, by walking lines and climbing poles, and analyzing data by hand. New surveying methods and AI-driven asset condition assessment solutions provide a better way to plan and analyze asset health. Digital, automated visual intelligence offers an unlimited view of different assets, asset types, asset components, and asset anomalies however deep in the service territory. Replacing land-based asset inspections with helicopters, satellites, drones or other aerial methods is faster, cheaper, and safer. Plus, comprehensive access to images, videos, thermal, hyperspectral and multispectral information can improve the accuracy of the system of record.
Maintenance, inspection and asset management are the largest OPEX budget lines for utilities. With automated asset inspection, the utility can immediately cut costs associated with traditional inspection approaches. Additionally, the powerful asset insights gained benefit change detection, storm damage assessment, and hotspot detection, all of which help the utility to detect and mitigate failures early and reduce maintenance and mobilization costs.
Now is the time for automated asset inspection, as it enables the insights utilities need to make data-driven decisions regarding which areas to focus on and prioritize. Rather than relying on manual observations recorded in the database against issues like corrosion or structural defects, a digital visual intelligence platform contextualizes all the information from different sources. Most importantly this improves safety and saves time by allowing linemen to prioritize their time making repairs instead of spending countless hours analyzing imagery to detect faults, and improves accuracy while classifying faults based on severity. No single company has significant market share in this space yet, leaving the niche open for the taking. Enter Buzz Solutions, MaC Venture Capital’s newest addition to the portfolio.
Buzz Solutions
Buzz Solutions’ platform is purpose built to detect faults in energy assets. They use a proprietary Machine Vision algorithm to detect major failure modes and anomalies on T&D lines. The Machine Vision model employs data gathered from various sources, including government agencies, nonprofits such as the Electrical Power Research Institutes (EPRI), power utilities, and third-party drone and helicopter service providers. This dataset consists of a large number of images of different electrical components of the T&D lines, such as conductors, poles, insulators, and towers. Their model is unique for this reason, as high volume of quality fault data is difficult to obtain.
Buzz Solutions was founded by two Stanford graduates who met through a class focused on solving problems and building startups within the energy sector. Through their research in the class, they interviewed over 35 power utilities, and discovered the massive opportunity in automating asset inspection in the industry. For the first two years they primarily focused on collecting quality data to train and build their algorithm, officially launching commercially in 2019. During this time, Kaitlyn Albertoli drove the effort as CEO, and Vikhyat Chaudhry remained CTO while also cutting his teeth working on AI projects at Cisco.
Vikyhat Chaudhary, CTO
Kaitlyn Albertoli, CEO
Buzz Solutions works with power utilities and third-party services to create a taxonomy to label and define the objects within these images. These images are then sent through pre-processing steps before being fed into the machine learning algorithm to remove “noise” from each image. This involves, for example, cropping and focusing on the object of interest in an image; resizing; changing angles of images taken from the air versus from the ground; and adjusting contrast and brightness for images taken under different lighting conditions. Through this process, Buzz creates datasets of the same electrical component taken from a variety of angles.
Their AI platform has become so robust through “active learning” which means the AI does a first pass and provides recommendations, and then humans in the loop penalize the algorithm for being wrong and reward it for being correct. This means as they continue to ingest more and more fault data, their algorithm will continue to become even more accurate. That being said, their algorithm already performs with 85% baseline accuracy, more than three times that of their closest competitor. Furthermore, Buzz’s real time data processing saves 6–8 months of manual data collection and ~50% of per image costs.
The team is already making a buzz in the utility industry — see here for the announcement of their deployment with New York Power Authority — and has great potential to expand into other sectors such as manufacturing and oil & gas. In the short term we’re excited about Buzz’s ability to save time and spend on utilities. However, we’re even more enthused about the downstream effects, such as improved safety for linemen, and reduction in damages such as wildfires that are caused by faulty assets, which not only cost utilities millions in damages, but negatively impact thousands of residents every year.
MaC Venture Capital is a seed-stage venture capital firm based in Los Angeles and Silicon Valley that invests in technology startups leveraging shifts in cultural trends and behaviors. The general partners represent diverse backgrounds in technology, business, politics, entertainment, and finance, allowing them to accelerate entrepreneurs on the verge of their breakthrough moment. The firm provides hands-on support crucial for building and scaling category-leading companies, including operations strategy, brand building, recruiting, sales development, and mission-critical introductions. MaC Venture Capital is the result of a merger between Cross Culture Ventures, co-founded by Marlon Nichols, and M Ventures, co-founded by Adrian Fenty, Michael Palank, and Charles D. King. Find MaC Venture Capital online at https://macventurecapital.com and @MaCVentureCap.
