Have you ever had to assemble anything from IKEA? Following those pictograms to a successful conclusion always seems to frustrate one and takes a lot longer than the estimated time stated; but what if you had a robot to do it for you? It may result in a smaller amount of tears shed and Allen keys thrown across the room. (Or maybe that is just my experience.)
There has been rapid progress in the field of robotics in the past few years and one team wanted to try to make something that could mimic the manual dexterity of a human. Until recently, autonomous manipulation in robots has been rudimentary. Open scene to Nanyang Technical University in Singapore. Scientists there have developed a robot that can assemble an IKEA chair independently without interruption. You may think that this seems a fairly easy thing to execute (not me, obviously) but there are a lot of actions that would go into this task that had to be considered. It is one thing to do the same action over and over again but when you need to do a bunch of different things one after the other, this requires a lot more thought.
This robot made a “Stefan” chair, chosen because of its design for human assembly with no special provisions made for robotic assembly. The pieces of this chair were randomly placed around it so that not only did it have to assemble it but also had to figure out what pieces went where, at what point in time. It took the robot a mere 8 minutes and 55 seconds with an 11 minutes and 21 seconds prep time to plan the motion pathways and 3 seconds to locate the parts. I am not sure that I could compete with this speed; it would probably take me the 8 minutes alone to find the pieces; I feel inadequate.
This robot was designed by Assistant Professor Pham Quang Cuong and his team from the university’s School of Mechanical and Aerospace Engineering and was comprised of a 3D camera, force sensors, and two robotic arms fitted with grippers to pick things up and manipulate them. Algorithms were coded using three different open-source libraries to assist the robot in its job of putting together this chair.
The robot begins by taking a picture of each part on the floor and then generates a map of where the pieces would go. The randomness of the pieces was used to replicate the action of humans unpacking the boxed chair and organizing the parts before assembly. Next, the robot plans its two-handed actions quickly without them running into each other or becoming entangled. The force regulation needed to hold onto and then assemble things without breaking was a little tricky for the scientists to accomplish. Industrial robots are great at positioning but not so great at regulating control. (Let’s just say, I would not give one a kitten to hold.) The force sensors located in the wrists helped to determine the necessary force needed to consistently and accurately detect the holes by sliding a wooden plug on the surfaces of the chair and performing tight insertions.
Since the success of this test, the team has been working with companies to apply this robotic manipulation to a wide range of tasks and industries such as glass bonding in the automotive industry and drilling holes in metal components in the airline industry. The possibilities are great and would not be too expensive. The whole assembly was produced using commercial off-the-shelf (COTS) industrial hardware.
The main drawback to be considered is: “Although all the steps were automatically planned and controlled, their sequence was hard-coded through a considerable engineering effort.”* So, you can buy the parts, but you will still need some big-brained IT guy to write the code for you to get your own IKEA chair-making robot. It may be a few years before you or I have our very own robot assistant.
–Janice Willson
Photo Source: Wikimedia Commons (composite)
