Mobile robot localization using a fuzzy ART world model

Abstract

Mobile robot localization is the problem of determining the robot's pose from sensor data. The ability to know their own location in the environment is an important capability for mobile robots. It is necessary for (and necessitates) map building, and for motion planning. The navigation architecture of, integrates a multiresolution grid world model, and a fuzzy ART (FART) based world model which is composed of a set of rectangular geometric primitives, or features. This paper presents a mobile robot localization method that is based on the integration of the fuzzy ART world model and a Kalman filter. This localization approach is original in that it employs a new type of geometric feature map in the context of mobile robot localization: the fuzzy ART geometric feature. This extends the previous navigation architecture, and demonstrates the application of the FART for mobile robot localization. The mathematical model of the localization algorithm is developed and applied for the case of a map composed of FART rectangular features. The paper presents experimental results that demonstrate the effectiveness of the proposed localization approach.