Mobile Robotic Manipulator Based Autonomous Warehouse Operations

Ørjan Øvsthus, Didrik Nicolai Robsrud, Lars Muggerud, José Amendola, Linga Reddy Cenkeramaddi, Ilya Tyapin, Ajit Jha

Abstract: The use of mobile robot-based warehouse automation systems is increasing with the larger logistic demands of modern society. To meet this requirement, the proposed method combines an unmanned ground vehicle (UGV), a mounted manipulator, and vision sensors for object detection and pose estimation.. The proposed method and associated functionalities such as - autonomous navigation, object recognition and pose estimation, leveraging manipulator for pick and place action is tested by creating a warehouse automation scenario at laboratory conditions. The associated functionalities are performed in modular and distributed fashion. While navigation is done in one computing platform and manipulator related task is done in another, both the computing platforms are connected such that the above functionalities are performed in tandem. The mobile robot successfully autonomously navigate from a start position to an arbitrary pick location within the SLAM-generated map in which the mobile robot locates itself. Then, it uses its manipulator and computer vision to detect a chosen object, estimate its pose and pick this object. Next, the robot autonomously navigates to a pre-defined place location within the SLAM-generated map and drop off the picked object. Finally, the robot autonomously navigate back to its pre-defined home position thus providing mobile autonomous manipulator towards autonomous warehouse operations. The detail implementation with source code for this work is available at https://github.com/orjano-max/ORJANOS_MASTER_REPO/tree/galactic and results of complete work at https://www.youtube.com/watch?v=hxrZh7bj16A.