California firm working with top EPCs to install six million solar panels in the next two years.
Robots are most effective in laboratories or controlled environments, such as a factory floor. When robots are required to automate tasks outdoors, the range of applications narrows with the number variables encountered. For this reason, construction sites remain labor-intensive locations resistant to robotics.
Solar construction sites may admit more automation than other sorts because of the regularity and essentially two-dimensional aspect of their layouts. Placement of heavy bases and even tracker tubes have been automated to some degree on many utility-scale solar projects for years. A key goal of these implementations is to have robots take over the more arduous tasks of materials transport and repetitive placement in all weather conditions.
More recently, robotics startups have been developing technologies for placing panels and even performing more dexterous finishing tasks. Some of these companies have received significant investment and contracts with engineering, procurement and construction (EPC) firms.
California-based Gritt Robotics focused squarely on large-scale solar construction for its approach to retrofitting commercial construction equipment with computer control featuring AI for improved flexibility and even site management support. Gritt recently announced it has deployed its robotic construction systems at solar job sites run by EPC Burns & McDonnell. In all, the company says it is it’s working with three top-10 solar EPCs to install six million solar panels in the next two years with a capacity of about 2.8 GW.
Puneet Puri, co-founder and CEO of Gritt, told pv magazine USA that large-scale solar sites have been an excellent proving ground for the company’s AI-assisted robotics technology.
“Outdoors is hard,” Puri said. “Outdoors, there’s going to be a lot of variability. Our thesis is that solar offers an example of a construction site where there are a lot of physically demanding, repetitive steps and materials transport that can be automated. At the same time, because it is outside, we can address aspects of a dynamic environment. We can build an intelligence layer that understands the challenges on the site, like trenches are open, people are moving through, materials may have snow or rain or dust on them, the light changes.”
Thus, large-scale solar construction has that right combination of regularity and variability where a robot with the proper common representation layer that can be trained for individual tasks under varying conditions. Ultimately, Gritt intends to apply the lessons learned from solar sites to data center construction and other infrastructure projects.
At the same time, Puri says solar has many repetitive tasks beyond heavy lifting that can be applied to other types of construction, such as tying rebar, applying fasteners, or attaching cables. It is all a matter of training. An aspect of Gritt’s software driven, AI learning approach is that nearly any type of construction vehicle or robotic arm can be retrofitted to perform specialized tasks autonomously.
“Caterpillar and Bobcat know how to build good vehicles,” Puri said. “For that matter, Kawasaki, FANUC, and KUKA know how to build good robots. Why should we reinvent the base platforms, which in many cases have been working in the harshest environments? We focus on writing the right brain to control them to do their jobs safely and repetitively.”
In terms of training, Vishal Dugar, co-founder and chief technology officer at Gritt, told pv magazine USA that like human workers, AI builds the ability to understand patterns, certain similarities of how to operate from one job site to another.
“When a person learns how to use a particular tool on one job, those skills remain constant,” Dugar said. “The AI also learns. It identifies all the similar patterns that it can reuse to make learning the next job that much faster.”
Puri and Dugar are alumni of the Robotics Institute at Carnegie Mellon University (CMU). Other members of Gritt’s leadership team also hail from CMU, or from MIT or other centers of automation research and practical application. There was a time when robots out in the field had to be purpose-built and shepherded by specialists to perform specific functions, whether descending into an Antarctic volcano or surveying the ruins of Chernobyl. Now, robots are on the cusp of taking on a wide variety of tasks guided by tutors rather than wrangled by technicians.
As indicated earlier, if the company views unstructured construction environments as the new frontier for physical AI, large-scale solar projects are proving to be an excellent proving grounds. It’s not just the relatively straightforward process of building out arrays. The industry is signaling a need for workers and warns of a growing shortage. According to recent industry reports, there is a shortfall of 53,000 workers needed to complete ongoing projects. Some estimate that 41% of construction workers nationwide will retire by 2031.
With competition for construction workers heating up, EPCs face a need to move and support personnel for projects in remote locations where accommodations are scarce. If this is the case for large-scale solar projects, it will also apply to data centers and other infrastructure projects facing opposition in populated areas.
“The mission of the company is to help civilization build infrastructure rapidly,” Puri said. “A big part of our vision is to help create the workforce of tomorrow. It’s not the same workforce that is doing the same things in the same way. It’s people who can use the tools we are developing to amplify what they’re able to do.”
This content is protected by copyright and may not be reused. If you want to cooperate with us and would like to reuse some of our content, please contact: [email protected].