Frustration and the torment of trying and failing to stack Ikea furniture may seem like a humiliation exercise to you, but know this: A nightmare of plywood one day can lead to robots that aren’t that stupid.
In recent years, robotics have been discovering that making Ikea furniture is actually a great way for robots to learn how to deal with real-world chaos. One group of researchers coded a simulator in which virtual robotic hands used trial and error assemble the chairs. Others, however, managed to procure a different set of robotic arms for the construction of Ikea chairs in the real world it took them 20 minutes. And now, a useful robot can help a man put together an Ikea bookshelf by predicting which part he will want next and handing it over.
“It’s one of those things that’s easy to try – even if we break a few shelves in the lab, it’s not a big deal,” said University of Southern California robotics engineer Stefanos Nikolaidis, co-author. recent paper describing the research, which was presented in May at the International Conference on Robotics and Automation. “It’s pretty cheap. And it’s also something we all have to do at some point in our lives. ”
Nikolaidis and his colleagues began by studying how different people build Ikea’s bookshelf. Instead of providing them with a pictograph instruction sheet, respondents were required to improvise the order in which they configured the support plates for the frame as well as the inserts for the shelves. (This is an important difference, because the bigger research question for this experiment is not about building furniture – more on that in a second.) Based on these results, researchers could group people into types or preferences. Some, for example, would attach all the shelves to one of the frames. Others would attach one shelf to each frame at a time. They are known as action sequences.
They then had to reassemble the subjects, this time with a robotic arm nearby to grab the pieces for them. The researcher would record which pieces (shelves or racks) the person started with, establishing a pattern for which the robot would guess. “Let’s say you go in and put the first shelf,” says Nikolaidis. “OK, the robot doesn’t know that much. Then choose another shelf. And now you start putting the third shelf. Well, it’s very likely that you belong to that group of users who put together all six shelves in a row. That’s very, very unlikely to then suddenly change your inclinations. “Once a robot understands a person’s preferences, it will give them the part they know that people like them have previously chosen the next one. Experiments have shown that this way the robot can adapt quickly and accurately to human style, successfully delivering the right components.
Think of it as a way for artificial intelligence researchers to develop an image recognition algorithm: if you want to detect cats, you feed neural networks with a multitude of images of cats. Having seen so many examples before, the algorithm can then generalize. If you show him a picture of a cat she has never seen before, she can rely on her previous knowledge to confirm that she is indeed parsing a hairy four-legged mammal with a fucking attitude.
This robot does the same thing, only instead of using a group of static images, it relies on examples sequence, the order in which people stacked shelves and racks, based on their preferences. “The robot knows that your next action will be handing over the next shelf with very, very high certainty,” Nikolaidis says.
In the end, however, this research is not about developing highly specialized robots that come to your home and help you make bookshelves. Nor is it about developing machines that can independently perform complex tasks like this. It’s about teaching robots how to work with humans without forcing them at all more crazy than people already get when they build Ikea furniture.
Despite all the blasphemy around robots coming to steal our jobs, the reality is that you are more likely to do so let the machine work with you him replace you immediately. For now – and probably quite a long time in the future – people will only be much better at certain tasks. No machine can replicate the dexterity of a human hand or come close to solving a problem like ours. What robots they are he is good at brutal business. Imagine a car assembly line: A robot puts a car door in its place, but small details require a human touch.