Development of an Autonomous Mobile Manipulator System for Fetch-and-Carry Tasks

Abstract

This paper presents a detailed development of a mobile manipulator to perform fetch-and-carry tasks autonomously in a structured indoor environment. This system was especially designed to prevent human contact in the work environment during Covid-19 pandemic situation. The system setup includes a Baxter manipulator mounted on a two-wheeled differential drive mobile robot. The ROS melodic platform installed on Nvidia Jetson TX1 embedded system was used for software development. Autonomous navigation and visual servoing are implemented using an Intel Realsense depth camera and a tracking camera which are collaborated with ROS Simultaneous Localization and Mapping (SLAM) and navigation. The system was validated under two experiments in order to: 1) analyze the performance of mobile manipulator navigation and 2) validate the overall system performance. Through several trials in each experiment, it was demonstrated that the developed system is able to perform fetch-and-carry tasks with the collision-free navigation consistently and repeatedly.