Abstract
Robot mapping and exploration tasks are crucial for many robotic applications and allow mobile robots to autonomously navigate in unknown environments. An accurate model of the environment is, therefore, essential for the robots to localize and perform navigation. In this paper, we present a line segment based mapping of indoor environments using range sensors for solving the simultaneous localization and mapping problem. The proposed method uses a modified Hough transform algorithm for line segment detection from laser range sensor data. The line extraction algorithm incorporates a noise model from the range sensor along with robot pose uncertainties. The proposed method is integrated with the extended Kalman filter. The extracted lines are merged to represent different structure in the environment correctly, and we show the results of our mapping method on simulated and real data sets. The experimental results demonstrate that the proposed method is capable of building an accurate line segment map of the environment for robot navigation.