As someone with a whole house full of IKEA furniture, this video immediately caught my attention: it features a robot assembling some of the Swedish company’s flat-packed furniture. (It also answers the age-old question, “How many people does it take to assemble an IKEA table?” The answer: two guys and a robot.)
Not only is the robot putting the table together, it’s actually learning the required motions based on physical input from its human handlers. This particular ‘bot was the subject of a research paper published in Proceedings of the AAAI Conference on Artificial Intelligence
In this case, the robot is learning to interact with its human counterparts in a “safe and natural way to accomplish tasks cooperatively.” The learning algorithms in use here include force and vision information, and the researchers have proposed a model in which humans can transfer “impedance-based behaviors” to a robot using kinesthetic teaching (touch-based learning).
Italian researchers Leonel Rozo and Sylvain Calinon developed the robot, which includes a force sensor and vision tracking system.
Good for that robot! It’s taken me several years to master the carpal tunnel-inducing hex-nut-and-allen-wrench art of IKEA assembly. Now if they could just build a robot that could sort through my pile of IKEA hex keys, I’d be set.
The next AAAI Conference will be held July 14-18 in Bellevue, WA.
Source: Programming By Demonstration