Seite - 61 - in Proceedings of the OAGM&ARW Joint Workshop - Vision, Automation and Robotics

A Robust and Flexible Software Architecture for Autonomous Robots in the Context of Industrie 4.0 Marco Wallner1, Clemens Mu¨hlbacher1, Gerald Steinbauer1, Sarah Haas2, Thomas Ulz2 and Jakob Chrysant Ludwiger3 Abstract—The next industrial revolution should allow the production of individually configured items at the cost of a currently mass-produced commodity. To make this possible, autonomous robots play an essential role. These robots use their knowledge about the world and the task as well as sensor information to derive the next action to achieve a common goal. To give research in this area the possibility to test novel methods and algorithms for the next industrial revolution, the RoboCup Logistic League was established. The league uses a small shop floor environment wherein a group of robots has to produce customized goods within a given time-frame. In this paper, we present a software framework for a group of autonomous robots which deal with the problems of the RoboCup Logistic League. The software is separated into three distinct layers allowing modularity as well as maintainability. The framework provides all the needed functionality starting from creating plans for a fleet of robots, to the hardware skills to detect machines, to move between locations and interact with the physical world. To show the use of the software framework a use-case is presented. This use-case is the exploration of an unknown factory hall with a fleet of autonomous robots. All the presented solutions in this paper were tested in the RoboCup world championship 2016 in Leipzig. There the system showed its robustness and its capability to solve issues arising with the next industrial revolution. I. INTRODUCTION To increase customer satisfaction and sales, the industry tries to fulfill the desire of the customer for an individual product to the price of a product created by mass-production. As current methods in industrial production cannot pro- vide these low costs for highly customized products, new approaches need to be developed. This is done through an ongoing automation in the industry which leads to the so-called Industrie 4.0 [1]. This term, originated by the German government, describes the abstract shape of the next generation in industry. One of the manifestations of this vision is the idea of a fully autonomous fabrication with smart machines. To create such a smart factory two parts are essential. On the one hand, there are configurable smart machines necessary which are capable of performing the manufactur- ing steps. On the other hand, an intelligent delivery system 1Marco Wallner, Clemens Mu¨hlbacher and Gerald Steinbauer are with the Institute for Software Tech- nology, Graz University of Technology, Graz, Austria. {mwallner,cmuehlba,steinbauer}@ist.tugraz.at. 2Sarah Haas and Thomas Ulz are with the Institute of Tech- nical Informatics, Graz University of Technology, Graz, Austria. {thomas.ulz,sarah.haas}@tugraz.at. 3Jakob Chrysant Ludwiger are with the Institute of Automa- tion and Control, Graz University of Technology, Graz, Austria. jakob.ludwiger@tugraz.at. between these devices is needed to allow the transportation of intermediate products to produce compound goods. As the usage of the machines and the scheduling can no longer be statically determined for a production line but need to adapt to the current requests, new algorithms need to be developed. To test these algorithms, the RoboCup Logistic League[2]was initiatedaspartof theannualRoboCupworld championship [3]. The idea is to create a simplified version of a smart factory which serves as a testbed and a standard- ized benchmark for novel algorithms and approaches. With this, different aspects can be tested and evaluated regarding distinct components as well as the complete framework. In thispaper,wepresent a software framework tosolve the challenges of the RoboCup Logistic League. The framework allows to schedule the entire fleet of robots and to react to changes in the environment in a reactive manner. Further- more, the system is capable of reacting to faults during the execution of a task assigned to a robot or even a full drop- out of a robot in the fleet. To allow an efficient, modular and maintainable implementation, the framework uses a layered architecture. This approach allows to schedule the fleet on the highest level while considering the reactive interaction between the robot and its environment on the lowest level. To show how this software framework is used we present in this paper how the robotic fleet can explore an unknown shop floor. With the help of this example, we will show how the fleet is scheduled to cover the shop floor efficiently. Additionally, we will demonstrate how the machines are detected and identified. The remainder of the paper is organized as follows. In the next section, we will discuss the RoboCup Logistic League in more detail. The proceeding section discusses the layered architecture. Afterward, we will discuss the software components which are used for the exploration of the shop floor environment. Before we conclude the paper we will discuss some related research. Finally, we conclude the paper and point out some future work. II. ROBOCUP LOGISTIC LEAGUE RoboCup, proposed and founded in 1997, is an annual international roboticscompetition.There, teamsfromallover the world compete in different disciplines, such as humanoid robots, soccer robots, rescue robots and the mentioned logis- tics competition. The logistics league simulates the problems arising in a smart factory. Primarily it provides a standardized testbed for test new algorithms and approaches for smart factories. 61