Evolving Neurocontrollers
|
|
A continuing project, most currently associated with the Texas ARP research on Intentional Systems. Originally research involved the evolution of neural models of hippocampal areas for cognitive mapping and goal directed navigation. I continue to explore pieces of the K-IV model of intentionality. More recently we have been looking at Liquid State Machines and Echo State Networks as models of statistical associative learning.
While these model simulate neuroscientific findings from biological brains, they are implemented as the control systems of real-world robots which have then been demonstrated to produce many psychological phenomena. This work is intended to demonstrate a new level of psychological theorising where biological implementation details are just as important as the peculiarities of our behaviour. This type of research has been called synthetic psychology, where the main aim is to synthesize complex, life-like behaviors using neural models on real autonomous robotic platforms.
Our main model for this synthetic psychology is the Khepera robot platform. We use both Khepera simulation in the Webots robot simulator, as well as physical Khepera robots. We obtained the Khepera hardware as part of the Texas ARP Intentional Systems grant in 2006.