A case study of a company that previously struggled with point cloud analysis provides an interesting example.
Imagine a connect-the-dots puzzle made of millions of dots (and in three dimensions). Now, imagine that getting the puzzle wrong might cost hundreds or thousands of dollars.
Until recently, that was the situation for architects and engineers performing point cloud scans of industrial spaces. They had to manually connect the dots of their scans while ensuring that the measurements were correct down to the millimeter, or risk providing a model that might cause their employers to make a costly mistake. The rise of partly or fully automated software to deal with the tedious mess of transforming a point cloud into a 3D CAD scan has made life far easier for industry professionals.
Brian Christiano, owner of BC Engineering and Design LLC, is one such professional. Christiano has a storied past, working as an engineer, an adjunct professor of nuclear and mechanical engineering, and a submarine officer for the U.S. Navy. These days, he works above the waterline, offering building information modeling (BIM), scanning, and visualization data of spaces to institutions like the commercial nuclear arena and the U.S. Department of Defense. But Christiano’s most popular service is laser scanning, offering 3D models of spaces.
He went into the field almost by accident: “I was working with the nuclear power industry and doing precision alignment. With one of the projects I was working on, a nuclear power plant here in Carolina, they asked me if I could model an entire room and determine travel paths in and out of that room,” Christiano explained. “And I ended up getting into the laser scanning arena.” But he’s a pro at it now—able to laser scan a room within hours, and replicate it exactly in digital form as a point cloud.
A point cloud is a digital record of a space, saved as a large number of points determined by color and position. The cloud is created by scanning the space with a portable 3D laser scanner. When the laser bounces off a point, its color and position are recorded. And when the scanner has cataloged every surface and object in a room, the result creates a point cloud, with millions of points from every visible surface in the room. You can think of it like a pointillist-style painting, in which lots of tiny dots come together to make a coherent-looking whole.
“It’s growing very rapidly as people realize that instead of taking a tape measure and a camera out to a site, they can bring somebody like me in and scan it, and accurately have whatever they need out of it,” Christiano said.
Point cloud scanning has become popular as a tool for estimating area dimensions and measurements, but there’s one major drawback to the process: the disconnected points remain just that. In traditional point-cloud analysis, a human worker needs to go through and painstakingly convert those points into a 3D mesh. And even before the technician converted those points, they would have to go through the process of combining the scans into a point cloud, lining up multiple different scans to create a whole picture.
Although the scanning process is relatively brief, the conversion into a model is a tedious, annoying one, especially since more recent scanners are capable of capturing millions of points per second.
BC Engineering had a particularly difficult time of it before adopting its latest software model. First, its primary analysis software only worked with the company’s FARO scanner, meaning that if the team members were collaborating with teams that used different scanners, they couldn’t import the other teams’ data. After they uploaded their data to the primary software, they would then have to go through five other software packages to analyze the data properly. Even worse, the software packages weren’t always compatible with each other, especially after one of them had received an update.
“One of the problems we were having is just the time to get it processed through to a CAD product. The other problem we were having is that [the programs] would lock up and crash. It was causing us time, it was causing us headaches, it was causing us frustration,” Christiano recounts. “We started discussing alternate software to process the data.”
The last straw for the company was when its cumbersome software meant it couldn’t complete a task for one of its clients. “One of the projects we were working on was for a nuclear power plant. We were doing the scanning and the modeling, and one of the aspects of the projects was to try to model a travel path with clash detection and things like that for a large heat exchanger,” Christiano said. “We weren’t in a position to define the travel path.” The company looked into a 30-day trial of Elysium’s InfiPoints, a “point cloud utilization” tool that allows users to process these clouds much more easily. After the trial, when Christiano realized the software could handle the company’s point clouds without the need for five other software packages, he bought it. The initial package cost approximately $25,000 and included several days of training.
According to Christiano, the product was well worth it. InfiPoints has allowed the company to extract shapes like pipes from the data in about 20 minutes, instead of painstakingly clicking on every piece of pipe to define it, a process Christiano said took “hours.” After the team got the new software, he looked at an old project, a point-cloud scan of a paper mill that he had initially scanned through FARO scene, Cubit and Cubit Point Sense Plant. The first time through, he’d had to manually extract many of the features, because the software didn’t pick up on features or didn’t detect certain colors. “We went back and did a comparison,” Christiano said. “We said, ‘let’s run this registered point cloud through InfiPoints and see what happens.’ It extracted all the pipe, it extracted all the facets and surfaces, and I went, ‘That would have been a lot easier.”‘
The company’s new software also lets the team import data from any brand of laser scanner, turn the points into a mesh, easily remove “noise” like people walking through the space, and perform virtual “flythrough” tours of the spaces. The last feature has been particularly useful when the team presents projects to clients. “For one of the plants we did a 90 percent completion presentation to the principals involved,” said Christiano. “We did a flythrough of the piping run. We started at the beginning, did a flythrough down as if you were Peter Pan flying down through the pipe.”
InfiPoints has been on the market since 2013. According to Danielle Perelli, customer engagement lead at Elysium, the purpose of the software “was basically to start bridging the point cloud world with the CAD world, because we had the CAD expertise.” Over the past five years, the market has grown, and both InfiPoints and competitors like 3DReshaper, PointCab and VRMesh are becoming more important to the construction industry. These programs are essential in helping industry professionals process the vast datasets contained in point cloud scans without having to use five different software packages or spend hours manually selecting shapes. They are making it easier to connect the dots, and in the process, making point cloud scans far less puzzling.