A sensor-based exploration algorithm for autonomous map generation on mobile robot using kinect

Abstract

This paper addresses on an autonomous exploration algorithm of unknown environment for mapping. An autonomous indoor carrying robot requires a kind of map as knowledge of the work space in order to efficiently execute the navigation task. SLAM (Simultaneously Localization and Mapping) is widely used as an generating a map. However, the map is based on the premise that humans operate robots equipped with sensors to generate the map. This is a burdensome task for the operator. Thus, an exploration method in an unknown environment for autonomously generating a map has been studied for decades. The main method is frontier-based exploration [1] which performs autonomous exploration using local map information. The algorithm is useful approach for the mapping unknown environments, on the other hand, the method has the following problem for applying to mapping method in a wide environment like a factory; since the map information which is being gradually extended is used, the calculation cost increases in proportion to the time [2]. From this problem, when the calculation time of the map information becomes long, the exploring efficiency will be deteriorated. Thus, it takes the time to generate the map. For more efficient exploration, an autonomous exploration algorithm using only infrared sensor and odometry information of a robot is proposed as a sensor-based exploration approach without using map information. The proposed method requires only a depth sensor and camera for which the Kinect is used as the equipment and one mobile robot. Turtlebot2 with Kinect is used as a wheeled mobile robot, and the effectiveness of the proposed method will be confirmed by comparing with frontier-based exploration and proposed method by indoor experiments.