VIEW-BASED COGNITIVE MAPPING AND PATH PLANNING

This article presents a scheme for learning a cognitive map of a maze from a sequence of views and movement decisions. The scheme is based o n an intermediate representation called the view graph, whose nodes co rrespond to the views whereas the labeled edges represent the movement s leading from one view to another. By means of a graph theoretical re construction method the view graph is shown to carry complete informat ion on the topological and directional structure of the maze. Path pla nning can be carried out directly in the view graph without actually p erforming this reconstruction. A neural network is presented that lear ns the view graph during a random exploration of the maze. it is based on, an unsupervised competitive learning rule translating temporal se quence (rather than similarity) of views into connectedness in She net work. The network uses ifs knowledge of the topological and directiona l structure of the maze to generate expectations about which views are likely to be encountered next, improving the view-recognition perform ance. Numerical simulations illustrate the network's ability for path planning and the recognition of views degraded by random noise. The re sults are compared to findings of behavioral neuroscience.