The innovation team at Swinerton, a major US-based general contractor, develops new methods and processes to solve construction’s oldest problems. Recently they began the process of automating progress tracking on work sites.
Today, the process is mostly manual. A project engineer, trade contractor, or other stakeholder walks the site and uses cameras or tape measures to record a rough estimate for the amount of work completed. However, industry-wide, these manual estimates are imprecise and subject to human error and bias, so the resulting figures are often inaccurate. They require multiple steps to verify, which in turn causes payment delays (often up to 90 days for final approval), cashflow limitations, and tension between project stakeholders.
In an effort to eliminate this cascade of problems at the source, Swinerton is developing a solution that automates the progress tracking workflow with the use of robotics, laser scanning, and AI processing. Swinerton’s vision is that this technology stack will gather hard data on completed work, which will speed up the payment process and ultimately improve the efficiency and profitability of every project.
As innovation analyst Tristen Magallanes explains, much of the technology for an automated progress tracking solution became available on the market as early as 2019.
It was at this time that Swinerton’s innovation team began experimenting with their first workflow. They started by sending a person through a job site carrying a Kaarta Stencil handheld laser scanner to capture highly accurate and comprehensive 3D scans of the as-built conditions.
With scans collected, the team would feed the data to Avvir’s software platform. The AI-powered solution would use algorithms to compare the 3D scans to the building information model, and this would enable granular, down-to-the-element tracking of work completed and time taken to finish, so the site managers could track progress against scheduling goals for individual trades. The product would also perform QA/QC and produce as-built deviation from design model reports.
At the end of this streamlined process, Swinerton would have all the hard data they needed to process payments.
However, this solution still failed to automate the most important part: moving the 3D scanner through the job site to capture the data. 3D scanning—even with a handheld device—still requires a significant amount of time, which can make it a difficult task for an engineer or trade contractor who is already spending long, busy days on site. This issue is only compounded when scanning frequently for progress tracking.
Magallanes knew that the innovation team needed to look to robotic platforms, which could carry the scanner, capture 3D data autonomously, and supply the missing piece of Swinerton’s end-to-end progress tracking solution.
She found the answer in Boston Dynamics’ Spot platform.
“I had tested a variety of wheeled robotic solutions that could go through different environments to collect data sets with different payloads,” Magallanes says. “Obviously, the biggest problem with those wheeled solutions on a construction site is that the environments are constantly changing and filled with obstacles they can’t get through. After testing probably two or three solutions, we determined that wheeled robots were just not going to work, ever.”
Quickly following that, she noticed that Boston Dynamics had opened up the early adopter program for its Spot robot and jumped to participate.
“Robotics is a technology that I had been tracking for a while,” she says. “I picked Spot because the platform exceeded other solutions we saw on the market. With its ability to move through a construction environment, and then, its ability to integrate, we could access the platform’s software and build with APIs if we needed to.”
She says that Swinerton was also attracted to Spot because her team wanted to work with a vendor that understood the needs of a construction company and the unique complexity of a job site. “We needed to work with people who had the breadth of services and capabilities to dive into this project and help us. We really needed a partner—which was a big selling point for us with Spot.”
After a period of development, Swinerton deployed the Spot platform to capture data at an active project in the Bay Area outside of San Francisco. The site spanned four floors, 197,800 square feet, and nearly 300 rooms.
With a Kaarta Stencil mounted on Spot, they manually planned a path for the robot at a high level, leaving the robot’s local navigation algorithms to account for the challenges of moving through different environments, climbing stairs, avoiding obstacles, and so on. Next, they prepared the space by opening doors, turning on lights, and removing a few hazards.
Then they began scanning. Spot navigated the complex construction environment to capture each floor—approximately 50,000 square feet—in 20-60 minutes. (The time per floor depended primarily on the amount of drywall that had been installed, the complexity of MEP work, and the segregation of spaces.) After capture, Swinerton fed the scans to Avvir and produced accurate progress tracking data.
Magallanes says Spot is key to achieving Swinerton’s vision. When the solution is finalized, she says, Spot will autonomously navigate the complex, obstacle-filled job environment, enable data capture with a minimum of input from the project teams on site, and feed the rest of Swinerton’s progress-tracking workflow.
The company estimates this automated process could reduce the time for final payment from 90 days to as little as 10 days, and eliminate the myriad downstream problems caused by slow payment processing.
But Magallanes notes that the benefits don’t stop there. For one, the robot has inspired the company to experiment with different kinds of data capture, and the innovation team is looking into using Spot to capture dimensional data for underground utilities or using the platform to capture environmental data as a method for monitoring safety in a work environment.
The implementation of Spot has also started to change the culture at Swinerton. It has pushed parts of the company to re-think large parts of their processes—and even refresh their approach to technology on the job site.
“Spot has brought a lot of good conversation about automating processes,” she says. “It drove different teams to be a part of the development, to investigate the tool and experiment with it, and help us solve problems. That’s one of the larger positive outcomes I didn’t expect.”
Swinerton is continuing to work with Boston Dynamics to achieve its long-term vision for Spot. Magallanes says the company hopes the robot will one day “live” on a job site, and automatically navigate areas at a regular schedule to collect various kinds of data as needed, with no human input at all.
The idea, she says, is that “someone from the VDC team will push a button from their desk and collect all the data they need. And then they’ll just process the data from their desktop.” Given Spot’s progress in the short time Swinerton has been working with Boston Dynamics, she says, that day will be here “sooner rather than later.”