A world-first Australian trial evaluates whether robotic solar panel installation can keep pace with rising panel sizes and tight construction timelines.
When Bouygues Construction Australia took on the development of the 500,000-panel 250 MW Goorambat Solar Farm in Victoria, it was keen to test whether robotic installation could support or accelerate large-scale solar delivery. As solar farms continue to scale to meet national energy targets, current manual installation methods are proving too slow and labour-intensive to keep pace.
And the industry’s increasing shift to larger, heavier modules, designed to boost generation, is making manual installation even less sustainable. Meanwhile, much of the installation workforce is made up of backpackers, with difficulty securing enough people for remote sites who can stick out a six-month contract.
Along with Bouygues, the Goorambat project attracted interest from European firm Equans, which had just finished building the Coleambally Solar Farm and was interested in understanding whether robotic technology could boost installation speed, improve safety outcomes and operate effectively.
The project required a robot that could integrate into existing workflows, adjust to site conditions, collaborate with crews and learn onsite.
As one of only three robotics companies worldwide automating processes on solar farms, Luminous Robotics Australia was contracted to deliver a 5000-panel one-block pilot, a self-contained section that could be directly compared with traditional manual installation.
Perfect fit
Luminous Robotics, a US-based company founded just two years ago by entrepreneur and robotics engineer Jay M Wong, was built on a philosophy that diverges from other robotics startups, said Tenzin Crouch, General Manager Australia.
“Jay’s thought process was … to actually go and find a problem to solve,” Crouch said.
Wong’s search for a real-world problem led him to volunteer on a solar farm. “He found the biggest challenge was the mechanical installation, particularly the manual labour of putting the panels in place by hand.”
This led to the development of Luminous Robotics and the fourth-generation, fully electric LUMI Series 4. Luminous Robotics Australia then launched in 2025 with support from Australian Renewable Energy Agency funding, and has now built, tested and deployed five robots across three different sites in six months.
Unlike some systems that require farms to be redesigned around the technology, Luminous aims for full compatibility: automating the most valuable tasks first, then redesigning systems for further automation.
“We specifically designed LUMI robots to fit into the exact workflow that all the construction companies use already.”
The LUMI robot automates the “pick and place” task, removing the most laborious and injury-prone aspect of solar construction while leaving fine-fastening tasks to human workers.
“A human will unbox the modules, then the robot will use computer vision to detect the position of that module,” Crouch said. “It will use its arm to come down, pick up that module, lift it up, detect where on the rail to place it, move the arm out and place it down, and repeat that process over and over.”
Manual installation typically requires a six-person crew. But with LUMI, only two people are needed, one to fasten and one to torque the bolts.
“A team of six is expected to install between 600 and 700 modules per shift. While the robot might only install 450 modules per shift, the number of people used on the job is less than half.”
The key metric the team was tracking was modules per person per hour.
“Towards the end of the deployment we reached around a 3.3-times efficiency improvement,” Crouch said. “The robot consistently installed 75 modules per hour with a crew of just two.”
Computer vision
With robotic installation, delays typically occur in small, incremental ways – for example, if unboxing takes longer than expected or when the robot needs to reattempt a placement. And the robot’s computer vision occasionally misdetects modules and rails, requiring retries and adding to the overall placement time.
“But we expect that over time, as the technology matures, we’ll get it up to nameplate capacity of 130 modules per hour.”
Solar farms are often built in challenging environments. With rolling terrains, changes in lighting and dust that can interfere with sensors. These environmental variables can’t be simulated perfectly in a lab.
“I can get a software release from my team in the US, put it on my robot, take it out to the farm – and something doesn’t work. Why? Because they tested indoors, or they didn’t have sloped terrain when they tested it,” Crouch said.
“We had some data from our initial projects in the US and used that to built out our initial algorithms for the different components of the robotic installation process: detecting the module and rail, and how to make the movement between the two.”
Critical insights
A critical element of the Goorambat deployment was gaining insights to improve efficiencies, including simple but high-impact learnings such as module staging.
“We decided that to lay out all the modules in the arrangement we theoretically planned was the best approach. But when we got onsite, we found it actually wasn’t the most optimal.”
The crew had to rearrange all the staged pallets of panels, adding a two-day delay to a month-long project. That experience informed planning for Luminous’s next deployment, where smaller test layouts will precede full staging. “We’ve fed that learning into the next one,” Crouch said.
During the Goorambat deployment, the US engineering team issued 27 software upgrades in just one month, each addressing real-world edge cases discovered onsite.
“We had this continual improvement cycle on a day-to-day basis. We collect the data from the site and push that back into improvements in the software itself.”
Safety measures
At present, Luminous Robotics operates on a rental model, supplying a robot and a technician to each project. These technicians – typically mechatronics engineers or mechatronics-trained technicians – undergo internal certification before deployment.
For onsite workers, a Safe Work Method Statement was needed.
“For example, if there are two rows of trackers, the robot would be in the middle, picking and placing with the human on the other side of the tracker,” Crouch said. “So even if something did go wrong, there’s no way they’re going to get hit by the robot.”
A short familiarisation period was built into the schedule, with about two days allocated for the team to get up to speed.
“The familiarisation was focused on how to work with the robot safely. It’s not a brand-new task for them; it’s more about working with the robot’s movement instead of the crew that would be lifting the module.”
Given Luminous is still working with the first 10 units of its initial production model, there are likely further technical upgrades that will be made to the robots.
“But as we scale the company and as the technology matures, the goal is to train people to be operators, either onsite or fully remote,” Crouch said. “The idea is that you don’t need an engineering degree to operate them. We need to build the technology in a way that anyone can operate it, and it’s intuitive and easy to understand.”
Part of this work includes building a dedicated training program, which is expected to be developed within the next 12-18 months. But the process is already in the works.
“We have already hired one of the people we were working with onsite at Goorambat. While we’ll upskill him in some technical areas, he brings valuable skills across production.”
Full integration
While initial per-watt costs of robotic installation are on par with manual labour, the long-term vision extends far beyond a single robot performing a single task.
“We can see a very clear pathway to achieving our mission: a 10-times expansion of solar globally,” Crouch said.
Key to that outlook is automated module movement. Currently, skid steers deliver pallets to site, but Luminous plans to automate those logistics so robots can coordinate with each other.
“We are now on another site deploying 10,000 solar panels and we have five robots here in Australia that we are already operating.”
Another frontier is fastener redesign. The current clips, bolts and torque requirements aren’t built for robotic dexterity. So Luminous is collaborating with module and racking manufacturers to redesign fastening systems with automation in mind.
“Is there a push-pin approach?” Crouch said. “Is there a completely different system, even riveting instead of nuts and bolts? I estimate it will be a couple of years before that’s more widespread, but it’s something a lot of people are looking at.”
This story was originally published in the February 2026 edition of create with the headline “Rise of the robots”.
WATCH: What does the future hold for human-robot collaborations?
