A team of researchers has developed a learning algorithm, dubbed ARCHER (Augmented Reward Chained Regression), which allowed the iCub robot to learn archery to the point where it learned to get a bullseye in eight tries from 3.5 meters away.
Here at Geekosystem, we like to follow the things leading up to Judgment Day, just so we can say “Told you so!” right before the big metal robot crushes our skulls, so it’s no surprise we’re a little worried when we find out that a program exists that teaches robots to hit bullseyes with arrows in eight tries or less.
As for how the ARCHER algorithm works:
…the ARCHER algorithm is used to modulate and coordinate the motion of the two hands, while an inverse kinematics controller is used for the motion of the arms. After every rollout, the image processing part recognizes automatically where the arrow hits the target which is then sent as feedback to the ARCHER algorithm.
Why does the program exist? Obviously, because the team of researchers who developed the algorithm never saw The Terminator.
(Petar Kormushev via Engadget)
