Robustness Evaluation on SLAM used in Petrochemical Complex Fires

Abstract

When using firefighter robots for fires in petrochemical complexes, there is a need to use Simultaneous Localization And Mapping (SLAM) for navigating the robot to the location of fire. Under this kind of severe environments, often times there is lack of necessary information for SLAM such as insufficient GPS signals, overheating of the LIDAR sensor, and the LIDAR sensor’s inability to observe fire. We have used Gazebo and ROS to simulate these environments in order to evaluate the accuracy of the trajectories estimated by our GPS and LIDAR-based SLAM module. We have found that the error becomes higher when the travelling speed of the robot is slower and that there is slight decrease in accuracy when the GPS signal availability is lower.