Scientist (03/14/05) Vol. 19, No. 5, P. 18; Spinney, Laura
Robot technology development is being increasingly influenced by physiology and neuroscience, and the time may come when robots will complement research in those disciplines. Current artificial intelligence efforts focus on imbuing robots with anthropomorphism, which is more likely to make machines capable of learning in a human-like manner and acquiring intelligence that can be applied practically. Giulio Sandini of the University of Genoa's Laboratory for Integrated Advanced Robotics will supervise a sweeping international project to develop RobotCub, an anthropomorphic learning machine designed to shed more light on human cognition and human-machine interaction. MIT, Delft University of Technology, and Cornell University made headlines last month with the debut of bipedal robots that walk using passive dynamics, but Rolf Pfeifer of the University of Zurich's Artificial Intelligence Laboratory says more effective walking robots can be realized by powering and directing motion in a more human-like way: He notes that "The brain doesn't control the trajectory of the joints; rather, it initiates that trajectory and controls the material properties of the muscles." Luciano Fadiga of the University of Ferrara believes equipping RobotCub with a system modeled after the "mirror neurons" that allegedly enable people to understand and attribute meaning to the actions of others will give it a mechanism for learning. Meanwhile, the University of Osaka's Minoru Asada is applying developmental psychology to robots by designing self-developing structures linked to or within artificial neural networks. This allows the robot to adapt to increasingly complicated tasks within its environment. Such breakthroughs are yielding new insights into human development and neurological disorders such as autism and hereditary cerebellar ataxia.
Click Here to View Full Article
(Access to this article is available to paid subscribers only.)
No comments:
Post a Comment