ROS-Based SLAM for a Gazebo-Simulated Mobile Robot in Image-Based 3D Model of Indoor Environment

At present, robot simulators have robust physics engine, high-quality graphics, convenient customer and graphical interfaces, that gives rich opportunities to substitute the real robot by its simulation model, providing the calculation of a robot locomotion by odometry and sensor data. This paper aims at describing a Gazebo simulation approach of simultaneous localization and mapping (SLAM) based on the Robotic operating system (ROS) for a simulated mobile robot with a system of two scanning lasers, which moves in a 3D model of realistic indoor environment. The image-based 3D model of a real room with obstacles was obtained by camera shots and reconstructed by Autodesk 123D Catch software with meshing in MeshLab software. We use the existing Gazebo simulation of the Willow Garage Personal Robot 2 (PR2) with its sensor system, which facilitates the simulation of robot locomotion and sensor measurements for SLAM and navigation tasks. The ROS-based SLAM approach applies Rao-Blackwellized particle filters and laser data to locate the PR2 robot in unknown environment and build a map. The Gazebo simulation of the PR2 robot locomotion, sensor data and SLAM algorithm is considered in details. The results qualitatively demonstrate the fidelity of the simulated 3D room with obstacles to the ROS-calculated map obtained from the robot laser system. It proves the feasibility of ROS-based SLAM of a Gazebo-simulated mobile robot to its usage in camera-based 3D model of a realistic indoor environment. This approach can be spread to further ROS-based robotic simulations with Gazebo, e.g. concerning a Russian android robot AR-601M.