Multi-resolution hybrid mapping for autonomous robots

Abstract

This thesis examines the fundamentals of some common robotic mapping techniques that will be used in this research work, such as Hough Transformation line extraction technique, the Inverse Sensor Model which is used for occupancy grid mapping and EKF-SLAM which is used for localization. It will also present the concept of HYbrid Metric Mapping (HYMMs) technique which the proposed Multi Resolution Hybrid Mapping (MRHM) is based on. This thesis goes on to discuss some technical improvements that can be made to existing robotic mapping techniques. Although the Inverse Sensor Model is able to model the environment accurately, there is a drawback which may mis-classify the occupancy status of the grid cells. The fundamentals of the Inverse Sensor Model will be examined and its main drawbacks are discussed. An alternative approach is then presented to provide a more accurate occupancy grid mapping algorithm. Also presented in this thesis a technique which is able to reduce the time needed to construct dense grid maps, the Multi Resolution Hybrid Mapping. The concept of MRHM, which uses lines as base landmarks to propagate corrections from the feature map to the grid map, is then presented. Finally this thesis presents the simulation and field experiment setups and the obtained results.