On the Problems of SLAM Simulation for Mobile Robots in the Arctic Conditions

摘要

Autonomous robots in the Arctic cover a number of strategically important tasks, including climate research, reconnaissance, transportation, material delivery, search and rescue. These goals require adapting standard navigation, localization and mapping algorithms to the harsh Arctic conditions, which do not allow their straightforward usage. The paper describes main problems of using simultaneous localization and mapping (SLAM) algorithms in the Arctic region and formulate requirements for the Arctic landscape simulator. With regard to these requirements we constructed Arctic terrains in Gazebo simulator, which implemented three of the eight proposed Arctic features, and studied behavior of ROS implementations of GMapping, Hector SLAM, ORB-SLAM2 and RTAB-Map SLAM algorithms within the obtained terrains.