SPAUN-V:Extending Visual Perception Capability Of The SPAUN Brain Simulation System

In recent years,a large number of scientists have been engaged in the field of brain simulation,putting forward a lot of important concepts such as production system and conducting a series of large-scale brain simulation projects such as the Blue Brain Project and SPAUN.Among them,SPAUN is one of the most influential brain simulation systems.However,the visual perception ability that SPAUN can achieve is very limited.It can only perceive simple conformity and digital information without the ability to sense the complex object information.In order to enhance SPAUN's perception ability,we modified the architecture of SPAUN and build SPAUN-V,a brain simulation system with enhanced visual perception ability.The enhanced system can also perceive more complex object information,such as a dog,a car and so on.We input images into the system,through the systematic interpretation,encoding,storage and decoding of the image information,and finally write the result through a simulated manipulator,which accomplish the "eye-brain-hand"information processing flow like humans.The main contributions of this thesis are as follows:1)We try to use two different visual perception methods,analyzing and comparing their experimental results,finding the best perception methods.2)Based on the architecture of SPAUN,we propose a new architecture that can be used to construct visual perception enhanced brain simulation system and we use it to construct SPAUN-V.3)We implement a recognition application using the SPAUN-V.Based on the recognition application,we design a new application that can classify digits and objects.