Simulation of Autonomous Mobile Robot for Warehouse Applications Using Gazebo and RViz

Abstract

In this paper an Autonomous mobile Robot (AMR) is simulated in a warehouse environment using the ROS2 framework. The simulation model was referenced from an existing design of a functional AMR project where the CAD files were simplified into base_link, wheels and LiDAR sensor. The modelling and modification of original CAD was done in Fusion 360 which also helped in the URDF conversion in the form of robot description. After organizing and modularizing the URDF file the robot was simulated in a warehouse environment referenced form AWS RoboMaker warehouse world, where robot was driven using a differential-drive plugin. A LiDAR plugin enabled the perception of the environment. Simultaneous Localization and Mapping (SLAM) was enabled using the SLAM Toolbox to generate a 2D occupancy grid map without any custom parameters. Path planning and autonomous navigation was possible through Nav2 stack, with global and local costmaps governing the planned path and obstacle avoidance. The simulated robot model successfully navigated to the user-defined goal positions when set via RViz, where map, transforms, laser scan, path and robot model were displayed. Obstacle avoidance was also demonstrated when dynamic obstacles were introduced into the environment. Visualization infers the viability of a fully simulation-driven AMR using open-source ROS2 tools, offering a inexpensive and scalable technique for analyzing navigation algorithms prior to hardware deployment.