Map building and localization on autonomous mobile robot using graph and fuzzy inference system

We propose a new localization and map-building method similar to that of human being in the navigation problem. Human being is assumed to solve a problem through four processes: exploration process (EP), decision process (DP), behavior process (BP) and learning process (LP). We call the processes as human being´s capability for solving problems. We try to solve navigation problems by transferring this human being´s capability into a mobile robot. Firstly, in the exploration process, the mobile robot collects the environment information and builds the map using the method called ´graph´. Secondly, in the decision process, the mobile robot selects a proper action on the basis of the percepted information and the generated ´graph´. Thirdly, in the behavior process, the selected action is implemented in accordance with the output of the FIS (fuzzy inference system). Finally, in the learning process, the parameters of ´graph´ are updated via repeated implementing. We show that this method is promising for the mobile navigation problems through a number of simulations.