Mobile Industrial Robotic Vehicles: Navigation With Visual SLAM Methodologies

Abstract

Mobile industrial robotic vehicles are using cutting edge technologies and have been widely accepted as a means of sustainability in the last decade. Recent navigation approaches are commonly divided into two categories (i) Laser-Based and (ii) Visual-Based. Many researchers proposed navigation systems for laser-based SLAM but their efforts both in the two-dimensional (2D) and the three-dimensional (3D) environments are still lacking critical information, such as color and texture, from the facility layout in contrast with visual-based methods. Moreover, visual-based methods use more affordable sensor devices, indicatively monocular, stereo and RGB-D cameras, that provide highly detailed information from the operation's environment. The reconstruction of the 3D digital twin environment is more accurate and detailed, enabling the mobile industrial robotic vehicle to navigate in the facility layout and accomplish a much greater variety of tasks. The proposed research discusses recent developments in Visual-Based methods and analyses various well-known proposed systems. Performance assessment is also reviewed using the Robot Operating System (ROS) to compare the discussed methods and discuss their suitability for various facility layouts.