It’s no secret that the combination of robots and artificial intelligence are changing the workplace as we know it. Over the past couple of years, we have seen robot chefs, robot bricklayers, robot lawyers, and any other number of artificial intelligence-enabled machines that can carry out the jobs once reserved for humans. One area we often view as relatively safe, however, are bespoke jobs — such as carrying out custom paint jobs in manufacturing. The reason for this is that, while industrial robots have long been able to help with processes like painting, this has previously been limited to large batch sizes of identical components — with humans carrying out the more complex paint jobs, as well as performing inspections where robots are used.
A new system called SelfPaint, developed by German and Swedish scientists from the Fraunhofer Institute, may change that. Instead of the comparatively dumb painting systems used in traditional assembly lines, SelfPaint is a self-programming system designed to reflect a world of user customization in which batch sizes for products is as likely to be one as it is 100,000.
SelfPaint starts its painting process by producing a three-dimensional scan of the component in question. From this, smart algorithms then simulate the trajectory of the paint particles to work out the optimum volume of paint and air that are necessary to achieve a desired coating thickness. The system then plans the robot’s path so that it can paint most effectively. After the painting is done, it then uses a beam of light at a wavelength between microwave and infrared to measure the wet, colored paint quality. All of this is achieved without human intervention.
It’s an impressive tech demo — and its creators claim that it can save up to 20 percent in paint, reduce solvent emissions by 20 percent, consume 15 percent less energy, and complete the work 5 percent faster than conventional painting processes. In other words, there goes another job once occupied by humans!
“The first half of the project has been completed,” researcher Oliver Tiedje told Digital Trends. “That means that the scanning, simulation, path-planning, painting, and film build measurement modules have been separately developed, while we hope that the interactions and interfaces will be developed by the start of 2019.”
Still, regardless of our fears about an AI job takeover, you cannot help but think SelfPaint would be a pretty sweet addition to any self-respecting “maker” studio. We await the inevitable Kickstarter in a couple of years!
