Evaluation of SLAM Methods for Small-Scale Autonomous Racing Vehicles

Simultaneous Localization and Mapping (SLAM) is a critical component of autonomous navigation, enabling mobile robots to construct maps while estimating their location. In this study, we compare the performance of SLAM Toolbox and Cartographer, two widely used 2D SLAM methods, by evaluating their ability to generate accurate maps for autonomous racing applications. The evaluation was conducted using real-world data collected from a RoboRacer vehicle equipped with a 2D laser scanner and capable of providing odometry, operating on a small test track. Both SLAM methods were tested offline. The resulting occupancy grid maps were analyzed using quantitative metrics and visualization tools to assess their quality and consistency. The evaluation was performed against ground truth data derived from an undistorted photograph of the racetrack.