In the present work, the analysis of the 4 state-of-the-art algorithms of visual-inertial SLAM is detailed and compared to each other: ORB-SLAM3, VINS-Fusion, RTAB-Map and DROID-SLAM. The base set used in the test, the TUM RGB-D Freiburg1_xyz benchmark data (along with its synchronized RGB, depth and ground-truth trajectory data) was the basis set. We ran the four algorithms using the same set of conditions and ran the Umeyama transformation in the case of rigid alignment. Then, we analyzed the resultant trajectories using Absolute Trajectory Error (ATE), Relative Pose Error (RPE), drift and scale consistency. The quantitative findings showed that the performance of the tested methods introduced is very variable among the algorithms of different samples. ORB-SLAM3 is the best localization with the ATE RMSE of 0.0091 m and RPE RMSE of 0.0385 m. The same but almost worse results were obtained with VINS-Fusion with an ATE_RMSE= 0.0115 m and RPE-RMSE=0.0458 m. The accuracy of RTAB-Map was slightly above a medium (ATE_RMSE=0.0192 m., RPE_RMSE=0.0778 m.). DROID-SLAM had the greatest errors (ATE_RMSE=0.0365 m, RPE_RMSE=0.1598 m), and a high bias in scale (scale ratio=14.99) that was great in drift (end drift=0.0325 m, segment drift=0.535 m.). Visual (geometric) simulations, in terms of cumulative distribution functions (CDFs), per-axis error plots, and radar (radar visual representations), showed that ORB-SLAM3 and VINS-Fusion exhibited superior trajectory stability, minimum drifts and more uniform scale maintenance as compared to the rest of the frameworks. But DROID-SLAM exhibited a large range of over-scaling as well as instability in motion estimation. In the final evaluation, ORB-SLAM3 showed the most optimal performance as well as the balanced performance of all the algorithms which were tested. Therefore, it is optimal with the real-time robotic navigation and mapping applications. VINS-Fusion is a solid option to use in the case of tightly coupled visual-inertial systems.