Simultaneous localization and mapping (SLAM) plays a vital role in the operation of autonomous robots. In a dynamic environment, the results of some commonly used SLAM algorithms may not be optimal. Therefore, Visual SLAM is used. This study presents a comparative study of SLAM algorithms for autonomous mobile robots in structured indoor environments, especially those found in warehouses. Nevertheless, GMapping and Cartographer SLAMs are preferred due to their ease of use, 2D LiDAR data, and low computational complexity. However, they are often confronted with problems of dynamic obstacles, loop closures, and localization drifts in large areas. To overcome these problems, this research aims to investigate the combination of Visual SLAM with an Intel RealSense D435i depth camera, YD LiDAR X4, and a custom wheeled mobile robot platform built with a Raspberry Pi 4B and four 100 RPM DC motors. This configuration is intended to facilitate the implementation and comparison of GMapping, Cartographer, and Visual SLAM algorithms under the same circumstances. The results of this experiment show that Visual SLAM greatly improves the localization accuracy, map quality, and loop closure detection in warehouse environments. These results support the application of Visual SLAM to improve autonomous robots’ navigation in real‐world indoor environments.
Kunchala et al. (Thu,) studied this question.