S-BPM supports the representation of processes using two levels: the individual subject behaviour level and the subject interaction level. These two levels are helpful when it becomes necessary to specify the individual subject's behaviour in di↵erent levels of details. We take a system-of-systems point of view and take a S-BPM specific look on interoperability. We use human-robot collaboration as a motivation and use-case to understand the interoperability requirements of adaptive system-of-systems. However, every type of human robot collaboration has di↵erent requirements not only for subject behaviour but also for the interaction level. We present the challenges and advantages of using SBPM in this setting.
Increased competition is pushing enterprises to become sustainable in economic,
environmental and social dimensions. To reach this goal, enterprises must
become true smart adaptive systems for being able to react rapidly and flexibly to
changing environments. This involves the human system of enterprises as well
as the technical. Networked Mobile devices, social networks and edge devices
produce great amounts of data. Smart decisions are required to make use of
that large amount of information. Next generation information systems need to
support the S3-Enterprise — Sensing, Smart and Sustainable Enterprise which
is a system of system where human and artificial systems collaborate [
Human Robot collaboration is one area which shows the high potential of
the S3 Enterprise. Sensors are required for understanding the user’s position
and activities. The robot is controlled by smart algorithms. For becoming
sustainable the collaboration needs to be adaptive, reacting to changing products,
production processes and customer demands. Since the human and the artificial
system are independent and loosely coupled, some coordination of the
activities is required. Both systems need an understanding what the counterparts are
doing [
The paper is organised as follows. First Enterprise interoperability is presented. The work described here supports modelling and use of systems which are loosely coupled. We take a look on S-BPM supporting interoperability in production-system-of-systems. As a motivation and specific use case, we use human-robot-collaboration to show di↵erent levels of interaction in productionsystem-of-systems. 2
Interoperability in general and interoperability on process level specifically is required to have support formation of a system-of-systems. In the following we argue about di↵erent levels of interoperability that need to be addressed in system-of-systems to maintain a loose coupling.
“Taking a system-of-systems (SoS) perspective [
In human robot collaboration, using a system-of-systems perspective, one
system is of human and the other of artificial kind. Di↵erent degrees of
integration and interoperability of these systems is possible from working independently
to dedicated and task specific enhancement of the workers cognitive or physical
capabilities [
An integrated system works seamlessly with other integrated systems. However,
when taking a system-of-systems (SoS) perspective [
Current EI&I research is following a system-of-systems approach [
“[I]ntegration is generally considered to go beyond mere interoperability to
involve some degree of functional dependence” [10, p. 731]. This dependence
implies less flexibility and less resilience since it combines the involved systems
in order to form a single whole [
Loose integration, also referred to as unified interoperability, is an approach
to support inter-operation of system-of-systems, where all systems share a
common meta-model [
A third, even more loose coupled approach to interoperability exists.
Federated interoperability describes systems of systems, where systems are capable
of negotiating interfaces and information-structures at runtime. Only a minimal
set of requirements is needed a-priori. This approach is sometimes called
latebinding and requires a semantic unification space where concepts that are used
by two or more systems can be mapped [
For example, domain ontologies, are used for semantic unification [
Orthogonal to the degree of coupling are the levels of concern where
interoperability barriers (may) occur [
Barriers of enterprise interoperability are discussed on business / organisational level, semantic / conceptual level, and technical / data level.
On business level interoperability issues stop tow or more enterprise systems doing business in general. Examples include incompatibilities in legal practice and country dependent laws. On the same level, when processes are not compatible supplier - customer relationships are deadlocked, where both enterprises wait for the other to perform an activity.
On semantic / conceptual level, models used in IT are incompatible. For example di↵erent granularity of models stop enterprises from calling others’ API to transmit data. Also conceptual barriers like di↵erent levels of granularity of exchanged information objects occur (Person objects modelled with a name property vs. objects modelled with a given-name and family-name).
At the technical level, enterprise interoperability barriers include among others, syntactic message formats, service interaction protocols and technical security aspects.
The European Interoperability Framework (EIF) [
Organisational semiotics, allows to discuss in more detail the relationship
between the model, its use and the addressed reality [
In addition to the structural requirements, EI&I is not a one-shot approach.
Enterprise integration and interoperability is a continuous process in a dynamic
environment. Having interoperability achieved, this state is easily lost due to
changing properties in the external environment or simple things like software
updates [
Overall Enterprise Integration and Interoperability is a multi dimensional system-of-systems research approach which facilitates flexibility and adaptability while still having a system for a specific purpose. EI&I is a model-driven approach, discussing not only the technical issues for exchanging raw data, but also the semantics of the models (and meta-models) and the pragmatic aspects in terms of what is triggered in organisations through the information exchange.
With respect to the system-of-systems approach interoperability on technical level, is a pre-requirement for systems to interact. Interoperability on semantic level is required to support a common understanding. However, only interoperability on process level assures that on organisational level tasks are enacted without disturbance. 2.3
Before going into more details how to support processes involving humans and
robots, we take a look at S-BPMN. S-BPM has been chosen, as it has a formal
basis [
S-BPM as an approach in general, and the available tools for S-BPM
support interoperability. The following Table 1 shows this support for enterprise
interoperability by the S-BPM framework itself [
This table shows interoperability on several levels. However, for human robot collaboration one (among other) interoperability gab exists between the human and artificial system.
S-BPM has also been used in the manufacturing industry, connecting the business process layer with the production process layer [33,24,34]. In that research a S-BPM Subject integrated using OPC UA technology the link to hardware and a production process which involves humans (as usual with S-BPM).
Three types of subjects have been identified in SO-Pc-Pro: (i) Service Subject (ii) Human Subject (iii) Coordination Subject
The Service Subject represents a certain functionality required in the production system. This includes for example the robots and sensors. However, the subject behaviour is pre-specified. This works fine for pre-defined functions, where Business Objects may be used to parametrise the machine.
The Human Subject represents the operator or worker. A user interface needs to be provided to communicate with the worker. “The main task of Coordination Subjects is to coordinate the interaction of the subjects in the production environment to achieve a certain goal. How this goal can be reached is basically defined by a set messages sent to di↵erent subjects. These messages then again trigger the execution of concrete task implementations. The coordination subject’s responsibility is to sequence these messages based on a defined flow sequence, current state of the system and the execution results.” [24, p. 3653]
In our approach this is the responsibility of the production process subject, which takes care of the controlling the interaction between production systems.
From a global perspective in the SO-Pc-Pro approach, the user’s workflow and activities are represented as internal behaviour in this approach.
With respect to the machine workflow, either some of the activities are refined to trigger OPC UA connected devices, or a subject is created that has the desired (internal) behaviour to connect the low-level machine control to the S-BPM process. The conceptual and organisational control flow of the former is from the user perspective. There is a single thread of control. From the interoperability considerations above, the work in SO-Pc-Pro [24] is an integrated approach.
S-BPM allows (to some extend) to support the coordination of activities in a system-of-systems involving human and artificial systems. It has a formal basis which supports activities for artificial systems and it is human readable supporting the human system. In the following we take a look at human-robot collaboration specific aspects. 3
So far we have argued that the manufacturing enterprise is a system-of-systems and support for interoperability (in contrast to integration) is required. This is particularly true on the process level.
In the following, Human-Robot collaboration serves as motivation and example to show the advantages and disadvantages of using S-BPM as interoperability support for system-of-systems involving humans and machines.
With respect to manufacturing, one of the most flexible production system are robots in general and human-robot collaboration more specifically in the context of this research paper. It therefore makes sense from a workflow and process perspective to take a closer look on interoperable and adaptive production systems-of-systems with special attention to robotics.
Figure 2 brings together three important aspects for adaptive production system-of-systems: (a) the production system in fig. 2 is modelled as a system-of-systems, and systems are process-oriented (i.e. for a certain step a specific process instance is implemented) (b) while the production system-of-systems exhibits the desired modularity and flexibility an overarching process is needed to coordinate the (independent) sub-processes of the systems. (c) S-BPM can be used to model systems and their internal processes. Interaction of systems takes place through messages containing business objects. In the following we present di↵erent degrees of human robot collaboration. The di↵erence is in the synchronization and coordination needs of the collaboration of human and robotic systems. First we present these collaboration scenarios, then we discuss technologies for implementation. 3.1
Human robot interaction can be understood in di↵erent degrees of physical coupling. It is the interaction and the concept used to synchronise activities that specifies the degree [35]: (i) A simple binary input (e.g. start / stop) from the worker to control the robot is a simple interaction. Point of synchronisation is through the interaction of the worker with the control switches in this simple case. (ii) Human Robot coexistence is a situation where both operate in close proximity but have no shared or synchronised tasks or work pieces. Here hardly synchronisation is necessary. Both must only make sure to not interfere with the others’ work, e.g. collision avoidance is required. (iii) Human Robot assistance is the situation, where the robot serves the worker without any active part or reasoning, simply obeying the commands of the user. Here the synchronisation between human and robot takes place through the command information transmission. (iv) Human robot cooperation describes a situation where the operator and the robot work on the same workpiece. Here the synchronisation takes place through the work piece. Both need to be aware where the other one works on the piece and does not take any steps that interfere with the other’s work. This requires an understanding of both about the others current and planed tasks. (v) The most intense interaction occurs when humans and robots share the same task. This situation is called Human Robot Collaboration. The synchronisation requirements here are not limited to the workpiece, but need to synchronise the activities. The timing and location of the worksteps are of importance. Also the upcoming activities of the collaborator. Both (the human and the robot) need a detailed understanding of the activities including their timing.
Figure 3 shows the the simplest form of interaction (i), the worker starts or stops the robot. Due to the simplicity of the interaction no subject behaviour diagram is shown.
Figure 4 shows the coexistence scenario (ii) where human and robot share the same workspace but no direct interaction is required. Here a monitoring subject triggers messages in order to warn about collisions. Often that is implemented in the robot itself. However, its possible to have that component as a separate module implementing the required functionality.
Figure 5 shows the assistance scenario (Iii) where human has full control over the robot.
In figure 6 both agents share the same workpiece (scenario iv & v). In the cooperation scenario (iv) the task lists need to be communicated to be able to understand if there is a conflict. A possible conflict would be that both occupy the same area on the workpiece. In advanced settings some planing algorithm might dynamically assign tasks to either party. In another scenario worker chooses a task which is communicated to the robot to then react. As mentioned above (v) in the truly collaborative scenario not only the workpiece is shared but also tasks. While the basic interaction will look similar to the cooperative scenario (iv), the level of detail is much higher. Not only the area on the workpiece, on which tasks take place but also the timing of the tasks needs to be synchronized. Hence, the interaction is time-sensitive. This time-sensitive synchronisation of tasks is only implicitly possible in S-BPM. 3.2
For integration on data and technical level, the OPC UA Standard has been used to integrate PLCs (Programmable Logic Controller) in Business Processes [24,33]. In this approach, information and process models are mapped manually to the OPC UA information model. Through a publish subscribe protocol the PLC receives / pushes information from / to the Business process. This explicit mapping also solves the problem of di↵erent information granularities required for information understandable by human and robot.
With respect to the types of human robot interaction, this approach could be used for binary interaction and co-existing interaction (scenario i, ii).
For the assistance scenario (iii) some other approaches exist, that allow to reprogram robots by workers on the fly [36,35]. However, in these approaches a process representation does not exist. It is assumed, that workers and robots are executing a simple sequence of tasks. Focus of that work is on the developments for dynamic reprogramming robots by workers. Hence, advanced workflows where for example parallel executed activities which need some synchronization can not be dealt with.
Figure 5. Scenario iii: Assistance - worker in command
In some other research technologies for planning and assigning tasks automatically to humans or robots is analysed [37]. Here also a elaborated process point of view is missing, to support more complex assignment and coordination of activities of di↵erent agents. This allows simple cooperation (scenario iv) based on simple task lists.
Missing is support for manual design and automated planning of advanced process structures where parts are executed by human and artificial agents. S-BPM as basis for human robot collaboration (scenario v) allows to support loose coupling through the two layer approach. Multiple aspects need to be further developed to reach this vision.
On the Subject Behaviour Layer the semantics of the di↵erent model elements
can be determined through an abstract state machine based formal model [
Another aspect for future research, is to enable automatic re-planing of processes and complex tasks using that ontology to support the detailed specification of robotic arm movements, articulate potential collision situation, specification of human tasks an the required resources, tools.
The third aspect for further research is a common S-BPM meta-model and and information model (e.g. upper ontology) forming a basis for federated interoperability where robot and human negotiate over task distributions on an ad-hoc basis. Federated interoperability also requires a shared meta-model supporting the understanding of systems that may assume several subject “roles”. The shared subject information contains (implicitly) interaction protocols, as subjects only interact with specific other subjects with well defined interaction (send / receive states and business objects exchanged).
This description needs to be shared through some infrastructure services, allowing agents to identify other agents and subjects. This directory service is a basic support for federated interoperability, where agents interact with a priory unknown other agents. 4
We have argued, that production is a system-of-systems. The systems have an individual purpose and a certain degree of autonomy. To keep this autonomy a coordination approach is needed that allows to maintain autonomy.
S-BPM, due to its two level approach of separating the subject-interaction from the individual system behaviour would allow to support process interoperability.
We have used human-robot collaboration as a motivation and use-case to understand the interoperability requirements of adaptive production system-ofsystems.
Advanced types of human-robot collaboration requires more information on available other systems, their capabilities, detailed understanding of the others’ tasks and activities. Semantic interoperability needs to be established for example using a common high level ontology describing business objects, tasks, processes and services that allow agents to discover that information.
To be able to have the robotic system execute a process, also automated planing needs to take place. As stated above, that in turn impacts planing and scheduling of the overall collaborative processes. Additionally planing and scheduling needs to be able to assign a certain task to a specific agent at runtime.
For highly interactive scenarios, the following is also missing but not discussed in detail: – time sensitive synchronization of tasks where human and robots have to execute their specific tasks concurrently – time constraints in general are missing, e.g. to stop activities after some time. – dynamic assignment of specific subjects that implement a certain subject interaction protocol
The research described in this paper has been partially funded by the European Union and the state of Upper Austria within the strategic economic and research program “Innovative Upper Austria 2020” and the projects ”Smart Factory Lab” and ”DigiManu”. 21. Bastarz, F., Halek, P.: smart4sense2act: A smart concept for systemic performance management. In Schmidt, W., ed.: S-BPM ONE - Learning by Doing - Doing by Learning. Volume 213 of Communications in Computer and Information Science.
Springer Berlin Heidelberg (2011) 109–114 22. Meyer, N., Feiner, T., Radmayr, M., Blei, D., Fleischmann, A.: Dynamic catenation and execution of cross organisational business processes - the jcpex! approach. In Fleischmann, A., Schmidt, W., Singer, R., Seese, D., eds.: Subject-Oriented Business Process Management. Volume 138 of Communications in Computer and Information Science. Springer, Berlin Heidelberg (2011) 84–105 23. Rothscha¨dl, T.: Ad-hoc adaption of subject-oriented business processes at runtime to support organizational learning. In Stary, C., ed.: S-BPM ONE - Scientific Research. Volume 104 of Lecture Notes in Business Information Processing. Springer Berlin Heidelberg (2012) 22–32 24. Neubauer, M., Krenn, F., Majoe, D., Stary, C.: Subject-orientation as design language for integration across organisational control layers. International Journal of Production Research 55 (2017) 3644–3656 25. Weichhart, G., Wachholder, D.: On the interoperability contributions of s-bpm.
In Nanopoulos, A., Schmidt, W., eds.: S-BPM ONE - Scientific Research. Volume 170 of LNBIP - Lecture Notes in Business Information Processing. Springer International Publishing (2014) 3–19 26. Kannengiesser, U., Oppl, S.: Business processes to touch: Engaging domain experts in process modelling. In Daniel, F., Zugal, S., eds.: BPM Demo Session 2015.
Volume 1418 of CEUR Workshop Proceedings., CEUR-WS.org (2015) 40–44 27. Oppl, S.: Subject-oriented elicitation of distributed business process knowledge. In Schmidt, W., ed.: International Conference on Subject-Oriented Business Process Management. Volume 213 of Communications in Computer and Information Science., Springer-Verlag Berlin Heidelberg, Springer-Verlag Berlin Heidelberg (2011) 16–33 28. Oppl, S., Stary, C.: Facilitating shared understanding of work situations using a tangible tabletop interface. Behaviour & Information Technology 33(6) (2014) 619–635 Published online: 02 Oct 2013. 29. Oppl, S.: Articulation of subject-oriented business process models. In: Proceedings of the 7th International Conference on Subject-Oriented Business Process Management. S-BPM ONE ’15, New York, NY, USA, ACM, ACM (2015) 2:1–2:11 30. Meyer, N., Radmayr, M., Heininger, R., Rothscha¨dl, T., Fleischmann, A.: Platform for managing and routing cross-organizational business processes on a network router. In Schmidt, W., ed.: S-BPM ONE - Learning by Doing - Doing by Learning. Volume 213 of Communications in Computer and Information Science (CCIS).
Springer, Berlin Heidelberg (2011) 175–189 31. Heininger, R.: Requirements for business process management systems supporting business process agility. In Oppl, S., Fleischmann, A., eds.: S-BPM ONE - Education and Industrial Developments. Volume 284 of Communications in Computer and Information Science. Springer, Berlin Heidelberg (2012) 168–180 32. Weichhart, G., Feiner, T., Stary, C.: Implementing organisational interoperability – the sudden approach. Computers in Industry 61(2) (2010) 152 – 160 33. Neubauer, M., Stary, C., eds.: S-BPM in the Production Industry. Springer International Publishing, Switzerland (2017) 34. Weichhart, G., Stary, C.: Interoperable process design in production systems.
In Debruyne, C., Panetto, H., Weichhart, G., Bollen, P., Ciuciu, I., Vidal, M.E., Meersman, R., eds.: On the Move to Meaningful Internet Systems: OTM 2017 Workshops. Number 10697 in Lecture Notes in Computer Science. Springer International, Cham, Switzerland (2018) 26–35 35. Pichler, A., Akkaladevi, S.C., Ikeda, M., Hofmann, M., Plasch, M., Wo¨gerer, C., Fritz, G.: Towards shared autonomy for robotic tasks in manufacturing. Procedia Manufacturing 11 (2017) 72–82 36. Matthaiakis, S.A., Dimoulas, K., Athanasatos, A., Mparis, K., Dimitrakopoulos, G., Gkournelos, C., Papavasileiou, A., Fousekis, N., Papanastasiou, S., Michalos, G., Angione, G., Makris, S.: Flexible programming tool enabling synergy between human and robot. Procedia Manufacturing 11 (2017) 431 – 440 27th International Conference on Flexible Automation and Intelligent Manufacturing, FAIM2017, 2730 June 2017, Modena, Italy. 37. Mu¨ller, R., Vette, M., Mailahn, O.: Process-oriented task assignment for assembly processes with human-robot interaction. Procedia CIRP 44 (2016) 210 – 215 6th CIRP Conference on Assembly Technologies and Systems (CATS).