Implementation and Evaluation of SLAM Systems for a Mobile Robot

For autonomous robotic systems, Simultaneous Localization and Mapping (SLAM) is often a necessity. In the last few years, several SLAM techniques have been successfully proposed for varying environments. Hence, it is important to measure the performance of these techniques depending on the use case. This paper presents a relative examination and benchmarking of mobile robot trajectories estimated by four ROS-based algorithms in an indoor environment: (a)lidar-based: Hector-SLAM and GMapping and (b)camera-based: RTAB-Map and ORB-SLAM2. A mobile robot was designed with sensors such as 2D lidar, encoders, inertial measurement unit, and depth camera. Thereafter, experiments were performed in an indoor environment, and information containing time-synchronized poses was gathered from sensors by teleoperating the robot on a testing track. Different SLAM systems were tested on this data and were compared with suitable metrics. The results demonstrate encouraging outcomes for lidar-based Hector-SLAM and camera-based ORB-SLAM2 algorithms.