Enhacement of 2D lidar-based SLAM in an accelerator housing using fiducial markers

Mobile robots are known for their flexibility and autonomy, being able to execute dif- ferent tasks in dynamic and non-structured environments. Regardless of the robot’s aim, the main thing underlying the correct execution of whatever task is their abil- ity to move autonomously in environments that might be known or not. For this reason, great attention has been paid to develop efficient algorithms of localization and mapping, in particular when it come to environments where the most common GPS-tracking systems can not be used. Nowadays, mapping the surrounding envi- ronment and localizing a robot inside the map created can be done simultaneously using SLAM (Simultaneous localization and mapping) algorithms. The aim of the thesis project is the implementation of a mapping and localization system inside an indoor repetitive environment. Due to the particular structure of the environment, navigation performances using standard localization algoritms such as Monte Carlo particle filter do not have optimal performances. For this reason the standard localization approach has been enforces using Arucos, a set of fiducial marker whose position is known. Tests have been performed using a real robot and the testing environment is the accelerator housing of SLAC (Stanford linear accel- erator laboratory). Thanks to the usage of the fiducial markers, safe and reliable navigation can be performed, avoiding collisions and obstacles.