Main Article Content
Simultaneous Localization and Mapping (SLAM) is technique that is used to perform mapping and identifying the position of oneself in an unknown or unfamiliar region. This technique is being extensively investigated for application in areas such as autonomous vehicles, robotics, virtual reality and augmented reality. This paper first provides an overview of SLAM technique and explores a few existing visual SLAM algorithms (ORBSLAM2, ORBSLAM3 and DynaSLAM). After this, the performance of these algorithms on benchmarking datasets such as KITTI, TUM and EUROC is analyzed by considering parameters such as absolute and relative pose error. The plot of ground truth and estimated trajectory for these algorithms are also shown in the results section of the paper. The analysis was done using a virtual machine running Ubuntu 20.04 in AWS. A tabulation containing the results obtained from the evaluation tool is presented in the results section of the paper.