Abstract

This thesis deals with the localization and map updating problems of mobile robots. We introduce three new fuzzy logic based localization and map updating algorithms for mobile robots that are equipped with a ring of sonar sensors. The first algorithm deals with the global localization problem, the second deals with detailed localization and location updating, and the third deals with map updating. Sonar sensors are inexpensive but inaccurate. To deal with sonar uncertainty, we introduce models for angular and radial uncertainty of sonar readings based on possibility theory, mainly possibility distributions (fuzzy sets). The concept of the fuzzy local composite map is introduced. This map utilizes the uncertainty models associated with sonar readings to give a description of the robot's surroundings. This map consists of a set of components, each of which represent the shortest distance between the robot and a detected object, and the orientation of the detected object with respect to the robot. In the proposed global localization algorithm, the constructed fuzzy local composite map is matched with the given global map to identify the robot's location. In cases where multiple locations are obtained, the robot moves to accumulate range information to be used to reduce the number of candidate locations. This process continues until a unique location is obtained. In the proposed detailed localization and map updating algorithms, two sets of fuzzy composite map components are identified. The first set includes the components that have matching counterparts in the global map and are used to update the robot's location. The second set includes the components that do not have matching components in the global map and are used to update the robot's global reap by adding new line segments