PARIS—When Airbus wanted to develop a smart machine or a robot with tons of artificial intelligence inside for a specific task, it used to as a matter of course turn to a startup. That takes one or two years. But this year, Airbus put into production smart machines it developed in about eight months with Akeoplus—to make parts for the A320, said Stéphane Morel, founder and CEO of Chateau Gaillard, France-based Akeoplus.
Morel developed “gateway edge computing” software called AkeoSpine to include computer science and machine learning in machines and robots. The software, he said, routinely decreases by three or four the time it takes to develop a smart machine.
Airbus by the end of this year will have three AkeoSpine-enabled machines running in production, Morel said.
Manufacturers need help from the likes of Akeoplus today, in part because they need to recover investments in machines in 12 months, not two or three years, as was customary in the past, he said.
“The second problem manufacturers have is that volume of production of products is really difficult to predict,” Morel said.
“For example, even in aeronautics, they have an idea of how many planes they are going to produce, but in reality it’s really hard to be sure that it will be two times the volume that they forecast,” he added. “So they push everybody in smart manufacturing to develop smart machines. A smart machine is also easy to reconfigure.
“Mass customization is clearly the biggest driver of this revolution,” he said. “Because now in production, we need to change very quickly.”
AkeoSpine can not only connect industrial robots from big makers like FANUC, Kuka and ABB to 3D cameras and sensors spread around factories but also include computer science on top of these robots. Akeoplus treats the robots like any other component. “This is a really interesting breakthrough that we bring to the market,” Morel said. “There is no place today in machines and robots to put computer science.”
In addition to selling its software, Akeoplus provides a service around helping companies select the right type of hardware needed to correctly run the software and transform robots that are otherwise “closed systems,” he said. ”Smart manufacturing is really new. This revolution is just starting. Many people don’t know how to start and how to develop smart machines. So we are a leader in France in this field.”
Akeoplus, which employs 30 people today and expects to employ 60 people 18 months out, is taking part in the first French accelerator “dedicated to the industry of the future in North America” to accelerate its visibility in the U.S., Canada and Mexico, Morel said, noting that his firm is targeting aerospace and defense, as well as automotive manufacturers.
“Because we are not a big company, it is better to be together with others, like a team, to be visible in the U.S. market—because the U.S. market is huge, and it’s moving very fast.”
Here in France, Akeoplus is also working with Thales, the maker of critical parts for the French military, to develop smart robots with many sensors and “a lot a lot of computer science inside,” Morel said.
“They can now reconfigure the machine for different satellite programs,” he said. “They can accelerate their production—and avoid very uncomfortable tasks for humans,” such as filling small holes with just the right amount of glue all day long. “A robot is much better than a human to do this. And at the same time, the machine can automatically track all the steps it took during the process,” whereas a human has to stop and record on a paper what he has done.
Akeoplus is also working with Safran.
For that French multinational aircraft engine, rocket engine, aerospace-component and defense firm, Akeoplus developed four smart robots capable of edge computing. Safran employs them to mark specific reference points, using letters and numbers, on critical engine parts.
“Today, most manufacturers use CNC machines to do this marking,” Morel said. “They are bigger, more expensive and not very edgy in terms of reconfiguration.”
Morel, 46, started Akeoplus—the name does not have a specific meaning—in 2006, when, he said, “in Lyon, France, the concept of the startup was not really known.”
He had worked for Volvo in Sweden, analyzing data from the sensors used in crash tests. And he chose to focus on helping manufacturers make sure their machines don’t crash—by using edge computing. To him, that meant helping them develop machines smart enough to accept APIs, such as deep learning algorithms.
Akeoplus has established tech partnerships with several robot makers, including Kuka, FANUC and Stäubli, he said.
Will those partnerships, along with its track record in manufacturing thus far, it be enough to go up against competitors like Cisco and Siemens?
Cisco “wants to develop products to help people to do smart machine,” so Morel is keeping an eye on it.
Siemens might soon “open their system to help people to bring smart algorithms to their PLCs,” he said. “But we are not sure.”