Having to assemble furniture from Ikea is an endless exercise in frustration, so how cool would it be to have robots that could do the job? Researchers from Nanyang Technological University in Singapore have now taken us a significant step closer to achieving that dream.
Robots are super fancy these days, with all their backflipping, rollerblading, and door-opening antics. But when it comes to performing tasks that require fine dexterity, machines still have some catching up to do. Delicate manipulation is hard for robots because it requires several skills, including visual acuity, tactile awareness, motion planning, force control, and multiple-hand coordination. What’s more, these skills have to be executed with care and often simultaneously.
As difficult as this may seem, however, new research published today in Science Robotics shows it’s possible to create such a system using commercially available, off-the-shelf robotic hardware. Researcher Francisco Suárez-Ruiz and his colleagues from Nanyang Technological University built a platform that managed to assemble an Ikea chair in about 20 minutes, and without having to mess with the robot’s factory settings. Not bad.
For the experiment, the researchers purchased a STEFAN chair kit from their friendly neighbourhood Ikea. Now, this wasn’t an excuse to sneak in some product placement; Ikea furniture is meant to be assembled by humans in their home, not on factory assembly lines where manufacturing happens in controlled and highly predictable environments.
NTU’s system consisted of commercially available hardware, including industrial robot arms, parallel grippers, force sensors mounted on robotic “wrists,” and a 3D camera. The point was to create a configuration that closely matched the human “hardware” package. For added realism, the researchers positioned each piece of the kit in random locations, which meant the bots weren’t pre-programmed to know the location of each part (sadly, the researchers didn’t replicate the complete Ikea experience by excluding some pieces from the kit).
The construction of the chair was performed in three phases: The robot had to identify and locate each piece in its area, plan its motions and avoid collisions, and decide how much force was required when attaching the workpieces. The robot used both visual and tactile cues for the operation, along with some heavy coding. In the future, advanced AI could be used to make the process more efficient, and to increase the adaptability of the robots.
The hope is to eventually use highly dextrous and adaptable robots for a wide range of applications, whether they be performed in traditional factory settings or in unpredictable environments, like a home or public space. The Ikea chair represented a fairly basic challenge, but we can expect more complex construction tasks in the future. This is yet another area where robots are poised to replace us puny humans—but when it comes to building Ikea furniture, they’re more than welcome to take the job. [Science Robotics]