Stereo Camera Integration for Autonomous Navigation and Visual SLAM in a Mobile Robot

Abstract

Navigation enables a mobile robot to be used in various applications, making SLAM fundamental in mobile robotics. However, traditional SLAM approaches fail to handle obstacles that are outside the addressed 2D plane, which can compromise navigation. VSLAM emerges as an option to extract more information from the environment, enabling the integration of visual information into navigation, encompassing the 3D plane. This work integrates a camera system to perform VSLAM in a real environment. Visual data is integrated to enable autonomous robot navigation. Results demonstrate that the system is capable of integrating environmental information into navigation, generating an incremental occupancy map more faithful to the region explored by the robot than a map generated by a 2D plane.