Self-programming robots have an element of science fiction about them, and maybe they will eventually take over the world. But for now, it’s benign welding robots that are causing sleepless nights for Forge Robotics founder Robert Cormican.
The young Galway man has developed a system that teaches welding robots to self-programme, reducing their set-up time from weeks to hours.
“Traditional systems are programmed by hand, point by point by point. This could mean up to 2,000 points for a complex part and it could take around 160 hours to complete,” Cormican explains.
“We can get that down to single digits by using cameras and sensors to map the part and AI to detect every weld possible and where the points could go.
“Our system is really easy to use,” he says. “The production/manufacturing manager simply selects the weld they want from a menu. They hit the ‘go’ button and the system programs itself. This saves weeks of manually dialling in every point while greatly improving accuracy and efficiency.”
[ Tesla boosts spending plan to $25bn in AI, robotics pushOpens in new window ]
Forge’s target market is the manufacturing industry, and to access his potential customer base at scale, Cormican felt he had to move to the US, where a major shortage of skilled tradesmen is putting the manufacturing sector there under huge pressure.
“Laser welding is a cutting-edge technology that’s up to 10 times faster than traditional welding processes,” he says. “It is facing massive adoption, yet there is a lack of skilled tradesmen and automation engineers in the US to programme these robots and keep them running.
“There is also a massive AI boom, but the infrastructure and skilled labour aren’t in place to build data centres and chip fabs fast enough to support AI. And I understand this problem from the inside,” says Cormican, who worked with Valeo Vision Systems and Analog Devices before leaving his job to lay the foundations for Forge Robotics at the Portershed innovation hub in Galway.
“As an engineer, I spent weeks tearing my hair out due to the cycle time, downtime, line yield and throughput issues I faced doing silicon development in the semiconductor manufacturing industry,” he says.
“Other systems monitor defects as the weld goes down, but this is a reactive solution. A defect detected is still a defect produced. Forge proactively modifies the weld program in real time, while the system also learns from any mistakes and doesn’t repeat them.”
In his early teens, having seen an electric skateboard on American TV, he decided to build his own. He got to know a local engineer who was working on battery technology and spent time working with him while he was still at school
At just 23 years old, Cormican is banking his future on the success of the Forge system, but he is not flying completely solo with the venture, as his company is being supported by the high-profile US-based Y Combinator start-up accelerator.
Y Combinator works its founders hard, but it also puts its money where its mouth is, with substantial investments in the companies it backs. Forge has received more than $700,000 (€610,000) in investment to date between angel investors and Y-Combinator funding of $500,000, as well as intensive mentoring from successful entrepreneurs, access to the extensive Y Combinator alumni network and the chance to pitch to investors.
“The US is a massive market with great opportunities for funding for the development of complex hardware and robotics systems. So, it made sense to start in the US market before expanding back into Europe and Ireland.
“The Irish Government’s new €250 million Enterprise Ireland seed and venture capital scheme is also a great step in the right direction to support Irish founders,” says Cormican, who is looking at launching a funding round later this year.
“Our initial focus is on Tier 1 manufacturers in the US,” he says, “companies that have multiple factories with R&D and innovation hubs designed for bringing early-stage technology onto the factory floor.”
Forge Robotics is also exploring strategic partnerships with laser welding vendors. “From there we will expand into smaller manufacturing sites in the US and into other markets such as Europe and Asia.
“There are a number of companies working on stand-alone robotic welding cells, but they focus on traditional rather than laser welding, and not much of their work is focused on augmenting the robots already on factory floors. We can retrofit,” adds Cormican, who is already thinking about extending the self-training system to other tasks currently done by robots, such as cutting and folding.
From a young age, Cormican liked building and taking things apart, and has always been driven by seeing how far he could push the limits.
In his early teens, having seen an electric skateboard on American TV, he decided to build his own. He got to know a local engineer who was working on battery technology and spent time working with him while he was still at school, so it was no surprise that a career in engineering beckoned.
“In third year [at University of Galway], I managed to impress my lecturer, Dr Brian Deegan, with my project, and he invited me into his lab to work on computer vision research. I was looking at how rain and light affect image quality, and how that affects AI performance in self-driving cars,” Cormican says.
“I was very lucky to get that experience, and from that came the opportunity to join Valeo Vision Systems, who are world leaders in self-driving car and camera technology. This gave me a really good grounding in vision systems and mapping robotics.
“I think the hardest part of starting the business so far has been the uncertainty,” says Cormican, who adds that 18-hour days seven days a week are currently the norm.
“The nature of an early-stage start-up is that you have to make a lot of assumptions to get off the ground. However, when dealing with early-stage cutting edge technology, I’d say that being able to adapt and pivot quickly is the best cure for the uncertainty.”
