2D Indoor Occupancy Grid Mapping for Mobile Robotic Platforms Using ROS-based ORB-SLAM2

• Main Authors: ALBAN NARANJO, RUBEN EMILIANO, Emiliano Alban
• Other Authors: CHEN, CHIN-SHENG
• Format: Others
• Language: en_US
• Published: 2019
• Online Access: http://ndltd.ncl.edu.tw/handle/wqh9qs

碩士 === 國立臺北科技大學 === 機械與自動化碩士外國學生專班 === 107 === This study covers a new frontier to ascertain whether Visual Simultaneous Localization and Mapping (vSLAM) has the ability, by itself, to perform further motion tasks as path planning and autonomous navigation by using a single RGB-D camera. The proposed solution is the creation in real-time of a standard Two-Dimensional Occupancy Grid Map (2D OGM) system for an indoor environment leveraging complete compatibility with the Robot Operating System (ROS). The robot pose is estimated by the state-of-the-art feature-based algorithm so-called Oriented FAST and Brief SLAM 2 (ORB-SLAM2). Subsequently, the map is built by the combination of the depth information and the pose of the mobile robotic platform (MRP). The experimental process is divided into two stages: virtual and real environments. The purpose of the first part is to collect data and gain expertise configuring the system, whereas the second stage delineates the possibility to implement the suggested approach in a real application. Therefore, the resulting map is compared with a ROS LIDAR-based map to validate the potential and system’s applicability utilizing ground truth maps to compute map-mapping metrics such as mean square pixel error and free/occupied cells ratio. Furthermore, this study aims to transmit essential knowledge to those roboticists who are beginning to delve a viable vSLAM solution to build standard 2D OGM’s.