Neuro-robot Vitroid - Living neuronal network with physical embodiment by a miniature moving robot

We developed neuro-robot system with closed-loop interaction between the circuit of living neurons and outer world interfaced by a miniature-moving robot. Using the system, an interface for interaction between neurons and outer world can be evaluated. We call the concept that LNC is the central processing unit of the neuro-robot “Vitroid”. As an example of interfacing algorithm between neuronal activity and robot behavior, we implemented SOM-based algorithm to generate robot behavior. The winner nodes of SOM were gradually gathered together by each category during a robot running, suggesting the mapping between inputted data and corresponding representative position in the clustering space is correctly updated during the robot behavior. In addition, SOM-Vitroid succeeded in generation of collision avoidance behavior and animal-like autonomous behavior.