The Neural Network Mechanism Of The Spatial Location

With the advancement of Internet technology,mobile devices with all kinds of navigation technologies arise at the historic moment.Meanwhile,with the development of biological technology,applying bionics to autonomous navigation equipment becomes the new tendency in the research of the navigation technique.With the analysis of experiment results,biologists prove that the mammals use the"brain GPS" system to position."brain GPS" system is mainly composed of two functional cells,namely place cells and grid cells.Place cells are able to remember locations of interest in the environment.Grid cells are able to provide to place cells a coordinate system.There are three main kinds for computational models of grid cells,They are Attractor-Network model,Self-Organization model and Oscillatory-Interference model respectively.The main objective of this study is about the Attractor-network model of grid cells.The research ideas start from simple,and then go to difficulty.First of all,from the perspective of the network model of existing pure grid cells,a simple model of fixed weight is built for proving the validity of the theory.Next,Hebb learning rule of the weights learning is introduced to improve robustness of the model.The author proposes the concept of conjunctive grid cells and builds a conjunctive grid cell model representing location and velocity information.Finally,when the position input is provided to the network,the accuracy in localization and the robustness of the model are improved.We increases weight to detect the noise and test the robustness of the proposed grid cells network.