CONCURRENT PATH CONSTRUCTION FOR GLOBAL MAP BUILDING IN UNKNOWN INDOOR ENVIRONMENTS

Abstract

The map building problem is the most fundemental issue in intelligent mobile robot reasearch. Several map construction methods have been proposed, but most of them require a predefined path before the construction of an environment map. In this study, a novel concurrent path construction algorithm for map building problem in a completely unknown indoor environment is presented. In the concurrent path construction task, the main idea is based on a lookahead strategy that estimates the next location where the robot will go, using its latest state and range readings. A local map is derived from the readings of 2D range-finders and line segments constituting the local map are also used to update a global map. A significant feature of this study is that it does not reqire a predefined path for navigation to construct an environment map. Moreover, optimum sensor readings are aimed to decrease the data processing time needed for a complete map construction by reducing the number of range readings from the surroundings. Simulation results on widely used and custom indoor architectures demonstrate that the proposed approach generates highly efficient paths and global maps for all handled problem instances.