November 19, 2020
3 min read
As I write this, our field team is running several robots on a production job in the Bay Area. We have completed more than half the floors of a 13-story residential building in San Francisco, and we’ve started in on a 20-story tower as well, along with completing handful of other projects. Our team has learned a lot over the past few months about what it takes to deploy robots in production on construction sites.
Setting expectations properly is always challenging in robotics, and even more so in the world of construction. In my robotics career, I’ve discovered that a lot of my job is to educate people about what is and isn’t possible with robots today. Everyone watches sci-fi shows and sees humanoid robots replacing people, while the reality is more akin to Roombas getting tangled in electrical cords.
[Side note: I am immensely grateful to the pioneering work of robotic vacuum companies, who have blazed a trail for robot adoption in consumer households. The fact that we can compare our robotic layout printer to a Roomba makes expectation-setting so much easier.]
So, back to our 13-story project. Up until this point, we had completed small projects, but had yet to take on a multi-floor job and be critical-path on the construction schedule. The project executive saw the potential in our product and wanted to give it a shot. But he, and his superiors, were concerned about taking a risk on unproven technology. Would the robot work? Would Dusty be able to complete the job on time? Layout is on the critical path. If we failed and couldn’t meet schedule, the entire project could be in jeopardy. But if we succeeded, robotic layout would be a game-changer for the entire company.
Our team did the best we could to prepare for the first floor. We had a fully-trained field team ready to put the robot through its paces. We checked and double-checked the CAD we would be printing on the floor. And we even brought a backup robot, just in case.
Nevertheless, when we showed up at the job site at 7am, our crew was dismayed. The floor was much more cluttered than any job we’d been on in the past. Some parts of the floor were inaccessible, behind safety barriers or buried under material stockpiles. Instead of the clear, open floorplan we had encountered on past jobs, we found stub-ups and conduits embedded in the concrete every few feet. Although we had tested our system with the occasional robot hazard such as this, we had never tested it on a floor with thousands of them.
Undaunted, our crew set to work. We set up in a corner of the floor and started printing lines. The project engineer and layout foreman stood off to the side, watching closely. More people came by, wanting to see what this new robot thing was all about. As soon as the first lines came out, measuring tapes were pulled out and accuracy checked according to the plans. The lines were spot on -- we had passed the first test.
But by the time we had completed the entire floor, our entire team was exhausted. It had taken us much longer than we expected, and we had been fighting with robot malfunctions and challenging site conditions the whole time. Nevertheless, our customer saw the potential in our product and wanted us to try again on the next floor.
Over the course of the next few floors, we continued to work with the customer to refine our process and streamline our productivity. We reoriented our engineering priorities around making this customer successful, focusing on features that would increase our field team’s productivity. At the same time, we worked with our customer to improve the site conditions to make it more robot-friendly, and eliminate unnecessary layout (e.g. the inner track line in duct shafts, or interior drywall finish locations on a multi-layer drywall finish).
By the time we hit the halfway mark on the building, we had settled into a groove. Our field team knew what they needed to do, and our productivity started to climb. As a data geek, I’m incredibly excited by the productivity numbers coming in from the field and watching them increase week over week.
From our customer’s perspective, they report that they’ve managed to save at least one day per floor in the schedule thanks to robotic layout. And because of the amount of detail we are laying down, conflicts between trades are being resolved sooner.
Buoyed by our success on this project and the others we’ve completed recently, we are starting to scale up. At this point we are currently completely booked through mid-January, but we are actively growing our field team and building more robots. If you have a project in our service area (50-mi radius around Mountain View, CA) which is starting early next year, we would love to talk to you!