Modular ROS Based Autonomous Mobile Industrial Robot System for Automated Intelligent Manufacturing Applications

The objective of this paper is to develop a modular Robot Operating System (ROS) based autonomous mobile industrial robot system for automated intelligent manufacturing applications. We propose a finite state machine based method to integrate and manage various modular functions on the mobile manipulator which makes it have a great talent on mobility and manipulation. For mobility, we perform simultaneous localization and mapping (SLAM) on our robot to build a map and locate itself. We also develop navigation and collision avoidance strategies, so that it can navigate to anywhere it plans in the indoor environment. For manipulation, we use an RGB-D camera on the end effector to perform eye-in-hand system which given an image and point cloud, leveraging the learning-based object detection and shape-based grasping pose estimation to achieve flexible manipulation capabilities. In the experiments, we have developed an Autonomous Mobile Industrial Robot (AMIR) with a carefully designed mechanism and a fully modular ROS environment in our robotics lab at NTU. In the experiments, we construct an intelligent industrial scenario: collecting materials and delivering products from one place to another. Using our ROS based architecture to control the data/command flow between functions, we have successfully completed the demonstration of the system and proved the stability of the system.