In the video (below), you can see Qbo interacting with its technician, who it can immediately recognize as Arturo from his voice using the open-source Julius speech recognition software. It responds to Arturo’s commands, identifies a penguin from an image, turns around, and “learns itself.”
Qbo’s ability to see is quite a lot more complicated, as it is equipped with stereoscopic vision that allows it to see and learn. It uses one of its eyes for “object recognition”, while the other performs “face recognition.”
When Qbo learns an object, it uses its eyes and the Speeded Up Robust Feature (SURF) to identify an image based on its features (i.e. end points, curves, and lines). After that, it assigns a word to the image and stores both into its internal knowledge database. So just like it was able to learn that a picture of the Linux mascot was a penguin, it assigned “myself” and “me” to flat image of its reflection.
While seeing ourselves in the mirror seems relatively simple to us, recognizing oneself, or what’s know as the mirror test, is a feat that even most living organisms are not able to accomplish–which is why your cat either freaks out or starts hissing at its reflection. Poor kitty.
The scientists’ next plan is to get see what will happen if a Qbo sees another Qbo in front of it. Hopefully it’ll be another successful breakthrough experiment. Or…
[The Corpra via Engadget]
Like this? You might also enjoy…
- This Is the First Full-Length Movie Filmed With a Smartphone
- Meet eSleeper, The Automated Arduino-Powered Kitteh Bed
- GeekBytes: Sending It to the (Mini) Printers
Get your GeekTech on: Twitter – Facebook – RSS | Tip us off