Vertical Integration and Service Orchestration for ... - Science Direct

Available online at www.sciencedirect.com

ScienceDirect Procedia Technology 26 (2016) 259 – 266

3rd International Conference on System-integrated Intelligence: New Challenges for Product and Production Engineering, SysInt 2016

Vertical Integration and Service Orchestration for Modular Production Systems using Business Process Models Sebastian Wredea,∗, Oliver Beyerb , Christoph Dreyera , Michael Wojtyneka , Jochen Steila a Research

Institute for Cognition and Robotics (CoR-Lab), Bielefeld University, Universitätsstr. 25, 33619 Bielefeld, Germany b HARTING Technology Group, Marienwerder Straße 3, 32339 Espelkamp, Germany

Abstract Individualized production challenges established manufacturing approaches in terms of modularization, flexibility and efficient reconfiguration. Modular production systems with high flexibility, capable of interacting with products and humans, and vertically integrated with the business environment provide new possibilities for overcoming these challenges. Such flexible manufacturing approaches require seamless integration between all involved parties from the enterprise resource planning level down to the shop floor. Service-oriented architectures with standards geared towards automation claim to provide easier development and integration of modular production systems. However, in order to support the desired flexibility and integrability, a coherent modeling approach for manufacturing processes is required, in particular focusing on the orchestration between production tasks, resources, humans and services. To address these challenges, we introduce in this contribution a system-level approach for flexible manufacturing comprised of reconfigurable modular production cells, which can be easily composed into production lines. Reconfigurability is achieved through exchangeable process components, the use of lightweight robot manipulators as versatile handling subsystems as well as a cell control scheme based on executable business process models to allow runtime service orchestration. The systemic approach can be applied to realize a customer specific production down to lot size one and was validated in a vertically integrated production line manufacturing an industrial product with high variability. The resulting system has been exhibited at different trade fairs and demonstrates our interpretation of the SmartFactory paradigm for individualized production. © 2016 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license c 2016 The Authors. Published by Elsevier Ltd. (http://creativecommons.org/licenses/by-nc-nd/4.0/). Peer-review underresponsibility responsibility organizing committee of SysInt Peer-review under of of thethe organizing committee of SysInt 20162016.

Keywords: Modular Production System; Business Process Model and Notation; Service-Oriented Architecture; Robotics

1. Introduction Individualized production challenges established manufacturing approaches in terms of modularization, flexibility and efficient reconfiguration [1]. Major challenges from a business perspective [2] include the needs to minimize the required amount of time from objective definition to start of production, to bridge the gap between enterprise and production levels, i.e. providing near real-time traceability and control of production processes as well as to ∗

Corresponding author. Tel.: +49-521-106-5148 ; fax: +49-521-106-89042. E-mail address: [email protected]

2212-0173 © 2016 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). Peer-review under responsibility of the organizing committee of SysInt 2016 doi:10.1016/j.protcy.2016.08.035

260

Sebastian Wrede et al. / Procedia Technology 26 (2016) 259 – 266

achieve a high level of automation across all involved stakeholders and systems. Hence, increasing flexibility while managing the intrinsic complexity of reconfigurable and vertically integrated production systems is a key issue for future production systems. The vision of a SmartFactory [3] was devised to provide solutions for these challenges. For instance, it shall be capable to efficiently derive production tasks from customer orders as well as to autonomously schedule the execution of the necessary manufacturing steps in a networked system of production modules providing the required manufacturing services. This contribution describes a step towards this vision by introducing (i) a system architecture that allows decentralized and parallel execution of manufacturing processes for customer-specific products in a production line composed of modular production cells, which are based on (ii) an approach to describe manufacturing processes using the Business Process Model and Notation (BPMN 2.0) as a standardized representation [4]. These models encode the required technical interaction between services exposed by the modular production cells such as PLClevel process component and robot activities, the product and its parts, just as manual actions executed by human workers. Furthermore, we explain (iii) the concept for modular production cells and the vertical integration in a (iv) demonstration scenario that we used to validate our approach in real-world applications. In the following, Section 2 introduces the related work in the context of modular production systems plus approaches to model the orchestration and coordination of manufacturing processes. Section 3 briefly describes the mechatronic architecture of the modular production cells, while Section 4 explains the overall system architecture by mapping it to the concepts of the Asset Integration Architecture [5] blueprint. Section 5 exemplifies the course of action in the system architecture when executing a customer order within the vertically integrated demonstration system. The paper concludes with a short summary and outlook to future work. 2. Related Work To cope with the requirements of individualized production, modern manufacturing systems need to dynamically adapt to change, must be easily customizable and configurable from standard parts [6]. One popular approach, which conceptually defines the basis for the work presented in this contribution, are Modular Production Systems (MPS) [7]. Within an MPS it is possible to adapt flexibly to varying demand by the dynamic addition or removal of modules from a production line. Prerequisites for the realization of such systems and the intended re-use are consistent modularization of hard- and software components [8]. Furthermore, versatile manufacturing components are required that support a broad range of manufacturing tasks. Within the system concept presented in the following sections, compliant robotic manipulators are used as versatile handling and assembly systems. Among others, the Rosetta project [9] demonstrated the utility of compliant robots in an industrial context. We utilize compliant robots in the proposed system to exploit their dexterity for assembly tasks and their safety features for close human-robot interaction, i.e. in order to ease the robot task programming in constrained environments [10] as required in the context of the modular production cells. Besides hardware modularity and versatility, the transformation of static tightly integrated production systems to flexible, rather loosely coupled modular production systems with service-oriented abstractions requires concepts for the design of the software architecture and the coordination of goal-oriented actions in these distributed manufacturing systems. Prominent recommendations on how to organize the layers of connected manufacturing systems are documented in the Reference Architecture Model of Industry 4.0 [11] or the Industrial Internet Reference Architecture [12]. Another less standards-oriented but nevertheless related approach is the Asset Integration Architecture [5]. For the orchestration and coordination of manufacturing processes in these architectures, dedicated modeling approaches are required that go beyond the functionality of the IEC-61131-3 languages, i.e. using dedicated graphical workflow modeling techniques such as GrafChart [13]. However, in the case of connected manufacturing in modular production systems, modeling languages that are standardized, support distributed execution and the orchestration between production tasks, resources, humans and services are required. Furthermore, vertical integration and alignment with business processes are major requirements. A language that supports these requirements, originally developed for description and automation of business processes, is the graphical modelling language Business Process Model and Notation (BPMN) [14]. It has already been used for visualization and alignment of manufacturing and business processes [2], for the modelling of processes and functions at the level of Manufacturing Execution Systems (MES) [15] as well as in the context of architectures for the Internet-of-Things [16].


Besides the aforementioned aspects, several related approaches report on research targeting flexible reconfiguration in the context of modular production cells. One such approach is an agent-based system which is fully implemented at the PLC level [17]. A major advantage of this approach is its real-time capability. Furthermore, the approach by Scheifele [18] and Friedrich [19] also uses standardized modules within a service-oriented architecture (SOA). A central control component, which is called Workflow Manager (WFM), organizes the production processes and handles the configuration of a module with the correspondent services. Naumann et al. [20] present a layered control architecture for robot cells supporting application-level plug-andproduce based on available services, automatic device configuration and setup of required device communication. This work focuses on eased specification of robot behavior but does not address the challenges of connected manufacturing and vertical integration of individual robot cells. The SmartFactoryKL [21] uses plug-and-produce methods for flexibly removing or adding modules during the plant operation. A PLC-integrated control system offers a SOA for data-consistent, secure and standardized controlling. A OPC-UA communication standard is used in this approach for vertical integration to provide process data through all levels of development and production.

3. Modular Production Cells Modular production systems shall support easy composition and configuration of product lines for novel production tasks on the basis of homogeneous manufacturing modules. The modular production cells developed in the FlexiMon project, cf. Figure 1, feature standardized mechanical, electrical and softwarerelated interfaces as a basis for easy composition. Furthermore, the FlexIMon cell concept utilizes a plug-and-play infrastructure at module level, which makes it possible to change or rebuild the topology of the production line very quickly by reassembling the modular production cells. The connection between several modules and the surrounding infrastructure is realized by a universal connector from the R Modular type series. For interHAN Fig. 1: An exemplary FlexiMon modular production cell. nal and external communication every manufacturing module features a standard Ethernet interface, while an industry-suitable field-bus is used for real-time capable communication. Each manufacturing module uses an embedded Soft-PLC control (Industrial PC that emulates PLC functionalities in software) that exposes services of installed components to the orchestration components. Additionally, functions are offered by the manufacturing module for discovering and integrating theses services during initial setup. Furthermore, the Soft-PLC realizes a safety concept for users on the basis of certified standard technology. Another standard component is an RFID reader which is used for identification of products and parts within the production flow. Currently, the material flow in and between the manufacturing cells is realized by a rotary conveyor belt with RFID encoded workpiece carriers. For human-machine-interaction every module is equipped with a touch-based user interface integrated in the module control system and orchestration architecture. This touch-based user interface supports users during operation, exchange and startup of components within the manufacturing cell and during maintenance tasks. To achieve the necessary flexibility for handling different workpieces, each module is equipped with a lightweight robot, cf. Figure 5.

261


IoT Cloud / M2M

Enterprise Applications

Backend

Gateway / Agent

ERP Order Control, Approval and Tracking b) Process Model

c)

OPC-UA Client

Material Flow

Virtual Line Control

c)

OPC-UA Client

OPC-UA Client

Module A

Module B

Module C

OPC-UA Server

OPC-UA Server

d)

Process Engine Adapter

Soft PLC

Soft PLC

Robot

Process Engine Adapter

Soft PLC

Visual Inspection

e) Frame Supply

Product RFID Tag

a)

OPC-UA Server

Adapter

CPS

OPC-UA Client

Process Engine

Process Engine

Cell Control

Devices

Shop Floor

Enterprise

262

Inset Supply

RFID

Robot

f)

RFID

Laser

Clipping

Robot

RFID

Conveyor Belt

Fig. 2: Central aspects of the FlexiMon architecture concept. Visualized along the lines of the IoT Asset Integration Architecture.

4. Integration and Orchestration Architecture As argued before, modularity and clear interfaces are required both in hard- and software components. In the case of connected manufacturing, the software aspects span different layers and implementation technologies, which are crucial for development, integration, networking and orchestration of software components and services on different abstraction layers. Figure 2 maps the core aspects of the system architecture elements to the concepts of the “Asset Integration Architecture” (AIA) [22]. AIA can be utilized as a blueprint for the description of vertically integrated IoT and Industry 4.0 systems. The individual production modules within the “shop floor” layer can be divided into three logical sub-layers (cf., Figure 2, module A-C). The specific sensor-actor components are collected in the “device” layer and provide production or product related manufacturing or analysis functionalities in form of encapsulated services. Together with their encapsulated services the specific sensor-actor components are denominated as “process components”. To fulfill real-time constraints the process components’ behavior is implemented either on a Soft-PLC with classical languages according to IEC 61131-3 or as a Cyber-Physical System with their own embedded controller. In addition, virtual components may implement purely virtual services executed in the runtime environment of the Soft-PLC or inside an embedded system (e.g. based on the HARTING IIC MICA platform). As a common core, all process components provide access to their functionality for the internal module workflow engine via specified service oriented interfaces or appropriate adapters. Those services and additional properties (communication endpoints, parametrization, HMI elements for the user interface, . . . ) as well as mapping of semantic interfaces onto BPMN concepts are described in a manifest, which is used as the basis for dynamic configuration and connection of software components. A complex process component is the integrated lightweight robot. The control of the robot is implemented on its own hardware with reusable basic functions (e.g. motion primitives with collision avoidance in Cartesian and joint space). Furthermore, these basic functions are integrated as reusable BPMN subprocesses (e.g. pick-and-place operations), which also can be transfered to other robot manipulators with standard kinematics. The functions necessary for using the servo gripper are triggered by the Soft-PLC via the fieldbus and also provided as a service. Orchestration of the services is performed by an extensible workflow engine (cf. “Cell Control” in Fig. 2). It serves as a central control unit for the operation within a modular production cell. The workflow engine interprets the BPMN-based process models that represent the entire product-specific manufacturing workflow for a module. The relevant BPMN concepts are being bound to the available components and services as well as the execution context by the self description. The self description includes tasks, signals and additional artifacts like data objects

263

Sebastian Wrede et al. / Procedia Technology 26 (2016) 259 – 266 UR5 Component

FlexiMon Engine g ▪

De Device-specific implementation of abstract robotics services Soft real-time control (125Hz) of robot manipulator Mapping and transformation of robot state to BPMN elements

▪ ▪

Robot Movement Planning ▪ ▪ ▪ ▪

Visualization Robot simulation Adaptive path planning Collision avoidance

AFAG Component ▪ ▪ ▪ ▪ ▪

Service orchestration Process modeling in BPMN2 Product-specific process instances

▪

IEC61131-3-based Implementation of abstract services for manipulation Real-time control of the servo gripper hardware Mapping and transformation of gripper state to BPMN elements

Fig. 3: Robotics subsystem exemplifying hybrid integration scheme. Due to the requirements of the AFAG servo gripper, it must be exposed as a service with an implementation in the PLC environment while the robot comes with its own controller, hence directly exposing its services to the FlexiMon engine. However, both components are jointly considered during movement planning.

needed during runtime. For communication between process instances and process components the extended BPMN2 engine (FlexiMon Engine) utilizes established protocols like OPC-UA, Beckhoff ADS or web services. If necessary, the process components must provide adapters to enable integration of special components (e.g. robot handling system). These adapters allow exchange of events, control instructions and general data. Furthermore special tasks, messages and events are used to represent the necessary user interactions and manual process steps. Those allow the modeling and active integration of the user into the process. The user is instructed about the relevant task by the user interface of the production cell. The virtual representation of the production line as well as the execution of the module overarching production process are conceptually positioned between the enterprise level and the production level as shown in Figure 2 (“IoT Cloud / M2M level” / Virtual Line Control). From a technical perspective, either a direct integration into existing ERP solutions is possible [14] or the BPMN process models that represent the overarching process control can be executed within an MES context [15]. Within the FlexiMon project overarching production processes are being managed outside the ERP systems by their own instances of the workflow engine, which can send specific requests to each module using special BPMN elements. In case of an error a decision is made on this level whether the complete work order needs to be aborted and restarted after rectification of the problem or whether the repetition of specific subtasks of that work order is sufficient. The workflow engine monitors the overall production progress and informs the ERP systems of the work order status via OPC-UA. Using this concept, it is possible to allow manufacturing of diverse products of batch sizes down to one on the same production line as long as the configured manufacturing modules can be configured to perform the necessary production tasks. Each module decides autonomously whether it is able to provide the necessary services for required production steps. Thus, for specific products not all modules of a production line might be involved in specific manufacturing processes. The FlexiMon architecture allows several modules to work autonomously on production processes without a central process control as long as a mapping of the workpiece to a parametrized process model exists. The modules are loosely coupled, as there are no technical dependencies aside from the transport system. Due to the continuous monitoring and logging of the process execution on all architectural levels, high traceability for the order-specific manufacturing process is achieved. Furthermore, this creates a basis for process analysis and optimization. Effectively, the instantiated BPMN processes represent a product memory for each individual work order that document and control the individual manufacturing process.

264


5. Vertical Integration for Individualized Production In order to demonstrate and validate the presented concepts under practical and economical aspects two different production lines were developed using real applications from the HARTING production, while considering user and customer requirements. Figure 6 shows a recent production line, which was exhibited on various fairs (i.e. Hannover R Messe 2015, SPS IPC Drives 2015). In its latest configuration the production line is designed to assemble an HANModular plug (see. Figure 4) according to individual customer requirements. In this scenario the customer specifies an individual configuration of the plug in the HARTING eShop. The individual configuration is comprised of choosing one of four frame lengths as well as the type and arrangement of 24 different inserts within the frame and customerspecific identification aspects for horizontal integration. The process components used in the individual modules range from removable magazines with active separators to components used for assembling safety pins and a laser-based marking system. The production process combines different pick & place tasks such as the precise insertion of variable inserts in the different frame types (see also Figure 5 for an impression of the robot tasks). It also involves further process steps such as marking the product with a QR code to identify the individual configuration of the client and an identifier for the integration the product in the value chain of the customer. Following the FlexiMon concept the production line was vertically integrated into an SAP system. After configuring the Han-ModularR plug in the HARTING eShop a job is generated in the SAP ERP System, which is processed via SAP Manufacturing Integration and Intelligence (SAP MII) and communicated via OPCR Fig. 4: Exemplary HAN-Modular Plug UA to the modular production line. The course of action in accordance with with customer-specific configuration and identity information. the presented architecture that eventually leads to the manufactured product is as follows (cf. Figure 2): a) The customer triggers the dispatching of a production order in the ERP system. b) The production order starts an appropriate order-specific process instance in the virtual line control, including the order data and relevant information necessary for the process steps. c) The defined production activities will be accepted for processing by appropriate production modules. At this stage the production modules know which order is connected to which identity and which modulespecific production process. d) Once the product on a workpiece carrier is identified at a production module by its RFID service and the module has successfully tested its own preconditions, the registered process instance starts the production process at the module level autonomously. e) The interpretation of the process instances by the workflow engine results in invoking services that are provided by the process components in the module. The production progress is stored and updated in the process instance. f) After the module-specific production process is completed, the product is passed on to the next available module that is capable of executing the next manufacturing step for the specific product via the conveyor belt.

Fig. 5: Robot operation within module workspace during a manufacturing task.

Fig. 6: Demonstrator of the modular production system at the SPS IPC Drives fair in November 2015.

The individual production steps are reported to the SAP MII, in order to monitor the progress of production of the plant. After all operations have been successfully completed on the chain level, the product will be reported back to the ERP system. Finally, the customer will be informed about the completion of its individual product via email.


The experiences during the FlexiMon project with the development of the production line demonstrators indicate that new processes can be implemented easier, faster and cheaper with a high re-usability of readily available modules. Within the modular ecosystem production lines can be assembled by existing standard components such as the cell and its base components, the robot and the transport system as well as available process components. Hence, the hypothesis is that over time only few specific components need to be engineered from scratch. Furthermore, the commissioning of the modular system for a new scenario including modeling of the underlying business processes was completed faster than with previous approaches, which was mainly possible due to the re-usability of hardware and software components as well as sub-processes. 6. Synopsis This contribution introduced an approach for flexible manufacturing comprised of reconfigurable production modules that can be easily composed into production lines. Reconfigurability is achieved through exchangeable process components, in particular the use of lightweight robot manipulators as versatile handling subsystems. A semiautonomous cell control based on executable process models described with BPMN2 is used to address service orchestration including human tasks and vertical integration requirements. The system can be used to realize a customer specific production down to lot size one and was validated in a vertically integrated production line automating the assembly of an industrial product with high variability. The resulting system has been exhibited at different trade shows and demonstrates a step towards the realization of the SmartFactory paradigm for individualized production. Future work will investigate whether the material transport between manufacturing cells can be carried out by the integrated robot manipulators to increase the flexibility of the production line topology as well as to reduce the effort required for the initial set up of the modular production line. Furthermore, we want to improve the self-description of the process components, i.e., by using and extending AutomationML [22] and to formalize the semantics of the domain-specific BPMN elements used in production process models. From a business perspective, we want to quantitatively assess the economic benefits of the presented approach for individualized production. Acknowledgements This work was funded by the German Federal Ministry of Education and Research (BMBF) within the LedingEdge Cluster “Intelligent Technical Systems OstWestfalenLippe” (it’s OWL) managed by the Project Management Agency Karlsruhe (PTKA). References [1] Brecher, C., Kozielski, S., Schapp, L.. Integrative Produktionstechnik für Hochlohnländer. acatech DISKUTIERT; Hannover, Germany: Springer Verlag; 2010. [2] Gerber, T., Theorin, A., Johnsson, C.. Towards a seamless integration between process modeling descriptions at business and production levels: Work in progress. Journal of Intelligent Manufacturing 2014;25(5):1089–1099. doi:10.1007/s10845-013-0754-x. [3] Bunse, B., Kagermann, H., Wahlster, W.. The SmartFactoryKL Perspective. Future Markets Industrie 40 2014;:26. [4] Object Management Group - OMG, . BPMN Specification - Business Process Model and Notation. 2015. URL: http://www.bpmn.org/. [5] Slama, D., Puhlmann, F., Morrish, J., Bhatnagar, R.. Enterprise IoT: Strategies and Best Practices for Connected Products and Services. O’Reilly UK Ltd.; 2015. ISBN 1491924837. [6] Nyhuis, P., Fronia, P., Pachow-Frauenhofer, J., Wulf, S.. Wandlungsfähige Produktionssysteme - Ergebnisse der BMBF-Vorstudie ”Wandlungsfähige Produktionssysteme”. Werkstattstechnik online 2009;99:205–210. [7] Koren, Y., Shpitalni, M.. Design of reconfigurable manufacturing systems. Journal of Manufacturing Systems 2010;29(4):130–141. URL: http://dx.doi.org/10.1016/j.jmsy.2011.01.001. doi:10.1016/j.jmsy.2011.01.001. [8] Vogel-Heuser, B., Kegel, G., Bender, K., Wucherer, K.. Global Information Architecture for Industrial Automation. Automatisierungstechnische Praxis (atp) 2009;1(51):108–115. URL: http://ojs.di-verlag.de/index.php/atp edition/article/viewFile/71/787. [9] Patel, R., Hedelind, M., Lozan-Villegas, P.. Enabling robots in small-part assembly lines: The ROSETTA approach - an industrial perspective. Proc of German Conf on Robotics (ROBOTIK) 2012;:1–5. [10] Wrede, S., Emmerich, C., Grünberg, R., Nordmann, A., Swadzba, A., Steil, J.. A User Study on Kinesthetic Teaching of Redundant Robots in Task and Configuration Space. Journal of Human-Robot Interaction 2013;2(1):56–81. URL: http://www.humanrobotinteraction.org/journal/index.php/HRI/article/view/98. doi:10.5898/JHRI.2.1.Wrede.

265

266


[11] Eple, U., Adolphs, P.. Referenzarchitekturmodell Industrie 4.0 (RAMI4.0). Tech. Rep. April; VDI/VDE Gesellschaft Mess- und Automatisierungstechnik; 2015. [12] Lin, S.W., Mellor, S., Miller, B., Durand, J., Crawford, M., Lembree, R.. Industrial Internet Reference Architecture. Tech. Rep. August; Industrial Internet Consortium; 2012. [13] Pauker, F., Ayatollahi, I., Kittl, B.. Service Orchestration for Flexible Manufacturing Systems using Sequential Functional Charts and OPC UA. In: International Conference on Innovative Technologies (IN-TECH). 2015,. [14] Chinosi, M., Trombetta, A.. BPMN: An introduction to the standard. Computer Standards & Interfaces 2012;34(1):124–134. URL: http://www.sciencedirect.com/science/article/pii/S0920548911000766. [15] Witsch, M., Vogel-Heuser, B.. Towards a formal specification framework for manufacturing execution systems. IEEE Transactions on Industrial Informatics 2012;8(2):311–320. doi:10.1109/TII.2012.2186585. [16] Meyer, S., Ruppen, A., Magerkurth, C.. Internet of Things-Aware Process Modeling: Integrating IoT Devices as Business Process Resources. In: Advanced Information Systems Engineering: 25th International Conference, CAiSE. Valencia, Spain: Springer Berlin Heidelberg. ISBN 978-3-642-38709-8; 2013, p. 84–98. [17] Schütz, D., Schraufstetter, M., Folmer, J., Gmeiner, T., Shea, K.. Highly Reconfigurable Production Systems Controlled by Real-Time Agents. In: 2011 IEEE 16th Conference on Emerging Technologies & Factory Automation (ETFA). ISBN 9781457700187; 2011,. [18] Scheifele, S., Friedrich, J., Lechler, A., Verl, A.. Flexible, Self-configuring Control System for a Modular Production System. Procedia Technology 2014;15:398–405. URL: http://www.sciencedirect.com/science/article/pii/S2212017314002096. doi:10.1016/j.protcy.2014.09.094. [19] Friedrich, J., Scheifele, S., Verl, A., Lechler, A.. Flexible and Modular Control and Manufacturing System. Procedia CIRP 2015;33:115–120. URL: http://linkinghub.elsevier.com/retrieve/pii/S2212827115006654. doi:10.1016/j.procir.2015.06.022. [20] Naumann, M., Wegener, K., Schraft, R.D.. Control Architecture for Robot Cells to Enable Plug’n’Produce. In: Proceedings 2007 IEEE International Conference on Robotics and Automation. April; Roma, Italy: IEEE. ISBN 1424406021; 2007, p. 287–292. URL: http://ieeexplore.ieee.org/lpdocs/epic03/wrapper.htm?arnumber=4209106. doi:10.1109/ROBOT.2007.363801. [21] Weyer, S., Schmitt, M., Ohmer, M., Gorecky, D.. Towards Industry 4.0 - Standardization as the crucial challenge for highly modular, multi-vendor production systems. IFAC-PapersOnLine 2015;48(3):579–584. URL: http://www.sciencedirect.com/science/article/pii/S2405896315003821. [22] Kuhlenkötter, B., Schyja, A., Hypki, A., Miegel, V.. Robot Workcell Simulation with AutomationML Support-An Element of the CAx-Tool Chain in Industrial Automation. In: Robotics (ISR). ISBN 9783800732739; 2010, p. 1076–1082. URL: http://ieeexplore.ieee.org/xpls/abs all.jsp?arnumber=5756920.