LEGO as an invaluable tool for learning not only applies to people, especially children, but also to machines. It has been proven, if only little by little, with research undertaken by various groups such as Autodesk. This team has been working on developing machine learning capabilities to improve industrial robots. Such industrial robots, like the ones found in car assembly lines, are usually seen as effective in putting things together. But it’s an open secret with tech researchers that most industrial robots, when encountering something not part of its programmed routine, will mess up their work. That’s what Autodesk’s “Brickbot” project is hoping to address.
The San Francisco-based research group assembled their Brickbot, consisting of two manipulator arms and a camera-sensor suite, to do something that would be effortless for a child but complex for a “smart” machine: sorting through components in a container then assembling them into a whole. Guess what the pieces used are. Well, that’s right – it’s LEGO. Click the thumbnail below to see the Brickbot in action.
Autodesk’s co-head researcher Yotto Koga explains the use of LEGO pieces for their machine learning development with Brickbot, saying, “By starting with plastic bricks, we’ve been able to keep the project manageable while still having the freedom to experiment from the design stage all the way to a finished product.”
With the promise being shown by the performance of Brickbot in discerning and putting together LEGO pieces, Autodesk is ready to see how this machine learning potential can be used in actual industrial robots, with future collaborations planned with a manufacturer and a construction outfit.