Posted on 22nd Aug 2012
At Meiji University, Professor Junichi Takeno and his reseach group have developed a conscious robot that identifies unknown events as new information to be learned. The robot is taught to perform a certain action upon seeing a red, green or blue color. When presented with a purple or other new color, the robot cognizes it and ties the acquired information to a new action. This action is repeated thereafter upon seeing that particular color. Previously, the robot malfunctioned because it could not recognize unknown events, i.e., information other than what was already learned. The new system is expected to be installed on a robot that operates autonomously by judging the environment by itself, for instance, in a disaster area where the entry of humans must be prohibited.
A recursive neural network called MoNAD, which basically consists of neural network modules, is installed in the robot’s system. The system incorporates a somatic loop and a representation loop to simulate the function of human consciousness by achieving consistency between cognition and behavior. The function of consciousness is created by arranging several MoNADs in a hierarchy. Each MoNAD can also function individually.
The conscious robot incorporates multiple MoNADs. While known information is processed quickly, an unknown event takes more time to be processed. This difference in processing velocity is used to detect an unknown event and capture it as new information. MoNADs are pre-assembled in an action list, and an action is assigned sequentially each time the robot recognizes a new color. Provided ample memory is available, there is no restriction on the amount of information that can be learned. Besides colors, shapes, sizes and other features can also be used as information.
According to Professor Takeno and his research group, the MoNADs satisfy most of the ten requirements of human consciousness advocated by the German philosopher E. Husserl. MoNADs also pass the mirror image cognition test. Mirror image cognition means that the image of the self in a mirror is cognized as the self. It is an index used to evaluate the high-level cognitive function of humans. Besides humans, chimpanzees, dolphins and some other animals are known to have passed the mirror image cognition test. For this reason, the system equipped with MoNADs is called a conscious robot.
Find out more on Prof. Takeno's latest book Creation of a Conscious Robot: Mirror Image Cognition and Self-Awareness.
Director and Publisher