Could a Robot Help Save Deteriorating Infrastructure?

Gecko Robotics is using state-of-the-art automation technologies to help repair critical facilities—many of which date back to the Cold War.

Infrastructure maintenance and renewal are becoming increasingly urgent issues as vital facilities such as power plants, refineries and manufacturing sites age. Engineers with the skillset to tackle these challenges will surely be in high demand.

One company, Gecko Robotics, has developed an innovative approach to infrastructure analysis and maintenance. “At Gecko, the mission of the company is to protect today’s most critical infrastructure and help give form to tomorrow’s,” says Jake Loosararian, chief executive officer and cofounder of Gecko Robotics.

A Gecko Robotics Toka 4 robot inspects a boiler wall. Image courtesy of Gecko Robotics.

A Gecko Robotics Toka 4 robot inspects a boiler wall. Image courtesy of Gecko Robotics.

Gecko builds robots that can climb walls, crawl across ceilings and navigate through pipes in some of today’s most important infrastructure: power plants, refineries, manufacturing sites, hydroelectric dams, and even ships and submarines. The robots analyze the assets to understand their structural integrity to anticipate when they’ll fail—and help prevent the failures from happening in the first place.

“We see governments and commercial leaders struggle with predicting failure within the physical assets that run sectors like power, manufacturing, oil and gas, municipal, aerospace and beyond,” says Loosararian. “Often, however, they’re relying on technology that was invented during the Cold War, and so the information they’re basing repair decisions on is inaccurate, or flat-out wrong. Customers need and want hardware-enabled software so they can predict weakness, failure, and repairs, moving them from a position of reactivity to one of proactivity.”

Using Robots to Detect and Prevent Infrastructure Failure

Gecko’s remote-controlled robots can climb vertically and horizontally, using magnets to adhere to a wide variety of materials. They use ultrasonic transducers, localization sensors, lasers, and HD cameras to scan for wear, corrosion, cracks, blistering and other forms of degradation. The robots also use localization technology to pinpoint exactly where the problem spots are located on the asset.

The data collected by the robots and Gecko’s software platform can produce an accurate, detailed and validated report within 24 hours. This enables facility operators to track wear and tear over time, predict when failures will occur, and plan out when to make necessary repairs—significantly reducing asset downtime and lost production.

Watch Gecko’s robots in action inspecting fuel silos.

Gathering Data Safely, Accurately and Quickly

In order to predict and prevent asset failures, the robots need to gather data. However, data collection in these sectors still often follows Cold War-era practices—which can be time-consuming, dangerous and inaccurate. A worker dangling at the end of a rope a hundred feet in the air with a sensor doesn’t collect much reliable data, is putting their life at risk and fundamentally can’t solve the problem.

Gecko identified the need to inspect infrastructure quickly and safely while also gathering much more data than conventional methods. The company’s solution: robots and software. The robots inspect the assets and create a data layer that overlays the physical layer—sometimes consisting of millions of data points. The software analyzes and interprets that data, identifying potential problem areas that can be repaired before they fail.

“What we’re doing with the robots funnels into the foundation for a data platform that allows for complete transparency and governance into the most critical infrastructure and ensures that we’re never surprised by a bridge collapsing or a hydroelectric dam failing or a power outage,” says Loosararian.

The robots are also able to monitor emissions and gather data to identify opportunities to reduce those emissions.

Using Advanced Manufacturing to Rapidly Develop Robots

Gecko used advanced manufacturing technologies to create its highly specialized robots.

The early models of the robot performed single point acoustic sensing and visual inspections on a payload. This robot was made up of a drive module that had neodymium rare earth magnets.

Gecko used rapid prototyping techniques, 3D printing, laser cutting, computer numerical control machining and other methods to quickly improve its robot model, enabling it to carry larger acoustic sensor payloads.

The robots’ drive modules have been designed to work independently and autonomously from each other so that it can navigate difficult terrain without losing its grip on the surface. The robots can even function underwater and factor in the different acoustic properties of water to collect accurate data.

“The sensing payloads on top of the chassis of the robots are specifically designed to have a lot of modularity to whatever the contours or the assets actually are,” says Loosararian. “You could be inside of a pipe, you could be on a flat wall, you could be 300 feet up in the air, and you have to ensure that you are able to collect good signals … but you also have to ensure that there’s good adhesion with the robot.”

The company also used machine learning to train the robots how to rapidly process and analyze acoustic signals, and trained them on existing known defects caused by factors such as corrosion, as there are hundreds of different corrosion mechanisms that could impact infrastructure. Gecko uses its software to create a digital twin of the asset, which includes representations of its health and integrity. Prediction algorithms were designed in-house, using data not only from the robots but also datasets from other customers and existing knowledge of physics, to interpret where, how and when the infrastructure could fail.

Much of Gecko’s rapid iteration has been done on location at the infrastructure the robots are deployed to analyze—a much different approach than the conventional way to design a robot. “I think the fallacy of most roboticists is we love to design things in controlled environments,” says Loosararian. “That was the inverse of the way that Gecko originally created robots. We created robots by creating something maybe not as beautiful, but functional. And then we tried those designs out at the customer site literally every day.”

Gecko’s approach relies on rapid prototyping. Identifying failures in the field, developing an update at the lab and getting the robot quickly back in the field helps spur improvements and drive problem-solving. This enables the company to not only quickly respond to a client’s needs—but also to adapt to changes in the market.

“At the beginning we use plastics or wood before we use metal just to test out mechanical design ideas before we actually fabricate anything with a more expensive material,” says Loosararian. “We have to ensure that we have extreme modularity with the platform so that we can change the configuration of the robot very quickly at the customer sites to ensure that we’re able to actually work in an environment that was completely not built for robots to exist in.”

Repairing and revitalizing infrastructure is becoming a priority around the world. In response, Gecko is using modern technologies to identify problems and facilitate solutions for aging infrastructure.

“Combining advanced predictive software with technology like robots and sensors is the way to modernize our infrastructure,” says Loosararian. “The most promising technologies are those that access dangerous, yet critical, places that the human cannot, and can collect vital data about infrastructure, buildings, bridges and other assets. The most promising new infrastructure will be built with many of these systems already installed. We are finding that engineers are passionate about finding solutions to real-world problems and are excited to jump into the hard work of bridging the gap to a clean energy future.”