Visual Simultaneous Localisation and Mapping (VSLAM) for Unstructured Environment

Visual Simultaneous Localization and Mapping (vSLAM) is an important area of computer vision and robotics that enables autonomous systems to use visual input to navigate and construct maps of unknown environments while simultaneously constructing detailed maps. This paper explores key principles, challenges, and advances in visual SLAM algorithms, including Extended Kalman Filter (EKF), FastSLAM, and GraphSLAM. Using the data collected from visual sensors such as cameras along with the measured distance values, robots achieve unparalleled perception and motion execution for autonomous navigation. The fusion of vision and mapping propels robots into uncharted territory, revolutionizing the future research in the domain of robotics. This article explains the concept of SLAM and its importance in modern robotics and augmented reality applications. We review the theoretical underpinnings of Visual SLAM, including camera geometry, feature extraction, and corresponding matching algorithms, which are essential for accurate pose estimation and map construction