Exploration for autonomous 3D voxel mapping of static indoor environments with depth cameras and 2D odometry

This thesis focuses on the development of an autonomous exploration and 3D mapping algorithm for a mobile service platform in an indoor environment. The designated robot platform is the KUKA omniRob additionally equipped with eight Time of Flight (ToF) cameras and the algorithm is developed and tested within the institute’s internal Mobile Robot Environment (MRE) simulator. The Simultaneous Localization and Mapping (SLAM) problem is solved by a metaview registration method, in which range measurements and their corresponding odometry estimate are combined together by the Iterative Closest Point (ICP) algorithm. The algorithm operates on the raw range measurement without any feature generation. The result of the mapping process is a 3D occupancy grid map used for the autonomous exploration tasks. An information gain and frontier-based exploration routine is developed, based on a preceding discussion and analysis of literature on the subject. The implemented strategy uses an iterative approach for exploring the complete working space of the robot. Therefore, a collision free, traversable and reachable space is defined. This thesis introduces a 2D exploration grid map as a projection of the 3D occupancy grid map, which serves as an input for the given path planning process. The experiments within the MRE simulation environment demonstrate the applicability of the developed algorithm and give a critical analysis of existing assets and drawbacks. An autonomously working exploration algorithm is developed in this thesis, which provides a framework for further extensions and improvements to facilitate future researches.