Why scientists love robots building Ikea furniture

Frustration and You may think that the severity of trying Ikea furniture together and failing is an exercise in humiliation, but know this: particle nightmares can one day lead to robots that aren’t so stupid.

In recent years, robotics have seen that building ikea furniture is really a great way to teach robots how to handle real-world chaos. A team of researchers coded a simulator that used trial and error in the arms of virtual robots put on chairs. Others, however, managed to get different robot arms to build Ikea chairs in the real world It took them 20 minutes. And now, an auxiliary robot can help humans assemble the Ikea library by predicting and delivering what parts they want in the future.

“It’s one of those things that’s easy to test, even if we break a couple of shelves in the lab, it’s not a big deal,” says robotist Stefanos Nikolaidis of the University of Southern California. the final role describing the research, presented at the International Conference on Robotics and Automation in May. “It’s pretty cheap. And it’s also something we all need to do at some point in our lives. ”

Nikolaidis and his colleagues began to study how Ikea builds a library. Instead of providing this worksheet with pictographs, the subjects had to improvise the order in which they set up the frame support tables and shelf inserts. (This is an important distinction, because the main research question for this experiment is not about building furniture, more information about it in a second.) Based on these results, researchers can classify people into types or preferences. Some would attach all the shelves to one of the frames, e.g. Others would attach a single shelf to the two frames together. These are known as action sequences.

The subjects then had to reassemble, this time with the robot arm close by to take the pieces for themselves. The researcher would record which people (parts or supports) were starting, establishing a model by which robots could be detected. “Let’s say you go in and put the shelf first,” Nikolaidis says. “Okay, the robot doesn’t know that much. Then you choose the second shelf. And now you start putting on the third shelf. Well, it’s very likely that you belong to that group of users who assembled six shelves in a row. Very, very nekez then suddenly you would change your desire. ”When a robot knows a person’s desire, it will give it the part that people like it chose before. Experiments showed that the robot adapts quickly and accurately to human style in that way, successfully providing the right components.

Think of AI researchers as developing an algorithm for recognizing images: if you want to detect cats, you feed the neural networks of feline images. As he has seen many examples before, then the algorithm can be generalized. If you show him a picture of a cat he has never seen, he can use his previous knowledge to confirm that a four-legged furry mammal is studying him with a bad attitude.

This robot is doing the same thing, instead of using a bank of static images, it draws some examples sequences, in order for humans to tie shelves and supports together according to their preferences. “The robot knows that the next action it should take is to give it the next shelf, very, very confidently,” says Nikolaidis.

In the end, though, this research isn’t about developing highly specialized robots that will come to your home and help build libraries. Nor is it the development of machines that can perform such complex tasks on their own. It’s about teaching robots to collaborate with humans without even guiding them more People are crazier than they get when they build Ikea furniture.

Despite all the fuss about robots getting to steal our work, the reality is likely have a machine work with you than replace directly. For now — and probably for a long time to come — people will be better at some tasks. No machine can repeat the skill of the human hand or approach it to solve problems like us. What a robot are good is raw work. Think of an automotive assembly line: the robot pushes the doors of the car arms into place, but for fine details it has to be touched by humans.