Simulation Framework for Mobile Robots in Planetary-Like Environments
This work provides a tool for rapid prototyping of perception and navigation approaches for mobile robots in planetary-like conditions, but it is incremental as it builds on existing simulation and SLAM technologies.
The authors developed a ROS Gazebo-based simulation framework to evaluate navigation and localization algorithms for a planetary-like robotic platform, testing various visual and LiDAR SLAM methods in a simulated Martian environment with ground truth data for performance assessment.
In this paper we present a simulation framework for the evaluation of the navigation and localization metrological performances of a robotic platform. The simulator, based on ROS (Robot Operating System) Gazebo, is targeted to a planetary-like research vehicle which allows to test various perception and navigation approaches for specific environment conditions. The possibility of simulating arbitrary sensor setups comprising cameras, LiDARs (Light Detection and Ranging) and IMUs makes Gazebo an excellent resource for rapid prototyping. In this work we evaluate a variety of open-source visual and LiDAR SLAM (Simultaneous Localization and Mapping) algorithms in a simulated Martian environment. Datasets are captured by driving the rover and recording sensors outputs as well as the ground truth for a precise performance evaluation.