Skip to main content

Article Category

Article available in the folowing languages:

AI-powered robots to speed up metal and composite part production

Cognitively enhanced robots have recently been tested in metal and composite parts production scenarios, and their performance could have a defining influence on the sector. The AI-powered, autonomous and collaborative robots speed up production time and even learn from experience.

Digital Economy
Industrial Technologies

The wide gap between current robots and their future, science fiction-inspired lineage can mostly be summarised in two words: cognitive computing. It’s the same barrier that today prevents most robots on production lines from performing anything other than repetitive operations. Breaking it could result in faster production time and reduced cost. The COROMA (Cognitively enhanced robot for flexible manufacturing of metal and composite parts) project has taken this challenge head on by developing a ‘cognitively enhanced’ robot specifically for metal and composite parts production. Capable of understanding its surrounding environment and learning from it, this robot makes use of its embedded reasoning and sensing capabilities to perform better than its counterparts – autonomously. “Our robot builds upon vision-based scene understanding techniques to navigate a cluttered workshop and find the parts that must be manufactured,” explains Asier Barrios, researcher at IDEKO and coordinator of COROMA. “It can work in cooperation with humans, tap into the same know-how of the manufacturing process and mimic how we learn by means of sensors and machine learning algorithms.” Let’s take the example of grinding – an essential step in many industrial processes by which materials are mechanically broken into fine granules. The COROMA robot completely understands how a grinding tool gets worn. It predicts when unwanted vibrations may occur based on previous experience, learns how some areas of certain part types are more prone to have defects, and inspects these first to save time.

Combining AI-based autonomy combinations

Asked about what he considers as the most innovative aspect of the project, Barrios points out the combination of different autonomy-increasing solutions based on artificial intelligence. “This is true both from the point of view of the manufacturing process and from that of the human-robot interaction. Our solution allows robots to deliver improved performance, increased productivity and better part quality, while at the same time providing more adaptability to changing environments and production needs. It also results in increased safety, mobility and scene understanding features,” he says. To test its robots in action, the project team carried out prototype demonstrations in operational environments and workshops replicating factory floors. They used real parts from the aerospace, naval and energy-generation sectors, and had the robots perform a variety tasks. “The robots ground parts of aircraft engines, deburred and ground tubes and racks for the storage of nuclear fuel and inspected steam generator nozzles using ultrasonic technology. They also sanded moulds to manufacture the hull of large glass-fibre boats and machined composite parts for boats and aircraft profiles,” Barrios explains. The benefits were outstanding. Thanks to its embedded AI and mechatronics, robot programming time was reduced by between 38 % and 98 %. In some scenarios, the total manufacturing time was reduced by up to 60 %, while the time required to set up the manufacturing of new parts was reduced by up to 85 %.

Wide market interest

COROMA was successfully completed in September 2019, but work has continued ever since. Some of the project results are already commercialised by project partners as stand-alone solutions, such as robotic hand grasping software and laser-based object localisation hardware. “Several project partners are working on the exploitation of the complete COROMA solution for the grinding of metal parts and sanding of large moulds for composites. Some third parties indeed would like to take these to the market as turnkey solutions. We also have some technologies still in their certification phase such as the safety module, as well as others being refined to be commercialised by third parties. These include the prediction of stable robotic machining, grinding tool wear learning, and lifelong learning about the most probable defect-generating features of metal and composite parts for non-destructive testing,” Barrios concludes.


COROMA, AI, robot, industry, metal, composite, autonomous, manufacturing

Discover other articles in the same domain of application