He's worked with Gerald Edelman on neuronal group models at the Neurosciences Institute of Rockefeller University and is currently working on autonomous robots. Some movies of the robot "Monad" are available at his website (see external link, below). The available movie sequences illustrate the robot learning from experience. The robot interacts with color-coded objects that provide either positive or negative value, as measured by the robot's sensory system. The robot comes to associate object color with the value of objects, providing a result similar to the classical biological example of associative learning by animals such as chickens.
For example, the Darwin V system was an example of a neuronal group model for which the connection strengths in the model neural circuits were specified not by a programmer but instead by the model's behavioral interaction with an environment, interactions that were facilitated by use of an autonomous behaving device rather than a pure software simulation. Interactions between the Darwin V system and physical objects resulted in development of a position-invariant recognition of objects. This emergent feature of the system then allowed a transition to more complex behavior from the more simple behaviors that were present initially. Sporns and his co-workers have since published (Science Vol. 291, pp. 599-600, 2001) an influential challenge to the artificial intelligence community to take on the task of constructing autonomous robotic systems that develop through cognitive stages in the way that human children do. They call this a "new direction for machine intelligence: autonomous mental development".
Related articles
Biologically-inspired_computing
External links
Olaf Sporns' faculty page (http://php.indiana.edu/~osporns/)
He's worked with Gerald Edelman on neuronal group models at the Neurosciences Institute of Rockefeller University and is currently working on autonomous robots.
Sporns and his co-workers have since published (Science Vol.
Feeds |
Contact