SAMBA – an architecture for adaptive cognitive control of distributed Cyber-Physical Production Systems based on its self-awareness

SAMBA – an architecture for adaptive cognitive control of distributed Cyber-Physical Production... ORIGINALARBEIT Elektrotechnik & Informationstechnik (2018) 135/3: 270–277. https://doi.org/10.1007/s00502-018-0614-7 SAMBA – an architecture for adaptive cognitive control of distributed Cyber-Physical Production Systems based on its self-awareness L. C. Siafara, H. Kholerdi, A. Bratukhin, N. Taherinejad, A. Jantsch Factories in Industry 4.0 are growing in complexity due to the incorporation of a large number of Cyber-Physical System (CPSs) which are logically and often physically distributed. Traditional monolithic control and monitoring structures are not able to address the increasing requirements regarding flexibility, operational time, and efficiency as well as resilience. Self-Aware health Monitoring and Bio-inspired coordination for distributed Automation systems (SAMBA) is a cognitive application architecture which processes information from the factory floor and interacts with the Manufacturing Execution System (MES) to enable automated control and supervision of decentralized CPSs. The proposed architecture increases the ability of the system to ensure the quality of the process by intelligently adapting to rapidly changing environments and conditions. Keywords: self-awareness; cognitive systems; dynamic clustering; autonomous collaborating objects; system health monitoring SAMBA – eine Architektur zur adaptiven kognitiven Kontrolle verteilter cyber-physischer Produktionssysteme basierend auf Self-Awareness. Industrie 4.0-Fabriken nehmen rasch an Komplexität zu aufgrund der Einbeziehung einer großen Anzahl von cyber-physischer Syste- me (CPS), die logisch und oft physisch verteilt sind. Traditionelle monolithische Kontrolle und Überwachungsstrukturen sind nicht in der Lage, den steigenden Anforderungen hinsichtlich Flexibilität, Betriebszeit und Effizienz sowie auch Belastbarkeit gerecht zu wer- den. „Self-Aware Health Monitoring and Bio-inspired coordination for distributed Automation systems“ (SAMBA) ist eine kognitive Anwendungsarchitektur, die Informationen von der Fabrik verarbeitet und mit dem Manufacturing Execution System (MES) zur auto- matisierten Kontrolle und Überwachung von dezentralen CPS interagiert. Die vorgeschlagene Architektur erhöht die Fähigkeit eines Systems, durch intelligente Anpassung an eine sich schnell verändernde Umgebung bzw. Bedingungen die Qualität des Prozesses zu gewährleisten. Schlüsselwörter: Self-Awareness; kognitive Systeme; dynamisches Clustering; Autonomous Collaborating Objects; System Health Monitoring Received January 30, 2018, accepted April 14, 2018, published online June 4, 2018 © The Author(s) 2018 1. Introduction optimization and a number of sophisticated cognitive architectures Highly efficient production requires high degrees of flexibility, adap- [5, 6]. Even though the developments of technology have improved tiveness, and responsiveness in order to achieve high quality and the robustness and resiliency of Cyber-Physical Production System versatility of manufacturing processes [1]. To reduce lead time, previ- (CPPSs) in comparison to conventional automation systems, new ous approaches have focused on rigid and deterministic automated expectations such as self-diagnosis and prognosis, self-repair, self- production environments, which minimize disturbances during op- discovery and self-configuration, predictability as well as safety [7] eration [2]. However, the increasing structural complexity of produc- are rising. We categorize these expectations into three different tion systems, due to the growing number of CPSs and distributed challenges. (i) The first challenge is the self-aware health monitoring heterogeneous components in the loop, decrease the determinis- which means self-observation and fault diagnosis. (ii) The second tic nature of production processes and require agile controls capa- challenge regards the complication of the decision-making process ble of prediction and timely reaction to disturbances [3]. To achieve flexibility while enhancing system performance, real-time informa- tion from the shop floor shall be integrated into the control system. Siafara, Lydia Chaido, TU Wien, Institute of Computer Technology, Gußhausstraße Moreover, the optimal reaction should be decided with respect to 27–29, 1040 Vienna, Austria; Kholerdi, Hedyeh, TU Wien, Institute of Computer Technology, Gußhausstraße 27–29, 1040 Vienna, Austria; Bratukhin, Aleksey, Danube the system goals, which may themselves change during the oper- University Krems, Center for Integrated Sensor Systems, Viktor Kaplan Straße 2 E, 2700 ation. To further assure quality, an important task is the continu- Wiener Neustadt, Austria; Taherinejad, Nima, TU Wien, Institute of Computer ous measurement of individually varying product properties in early Technology, Gußhausstraße 27–29, 1040 Vienna, Austria; Jantsch, Axel, TU Wien, stages [4]. In view of these requirements, researchers have proposed Institute of Computer Technology, Gußhausstraße 27–29, 1040 Vienna, Austria (E-mail: axel.jantsch@tuwien.ac.at) various methods of intelligent sensing, self-organization, and self- 270 heft 3.2018 The Author(s) e&i elektrotechnik und informationstechnik L. C. Siafara et al. SAMBA – an architecture for adaptive cognitive control... ORIGINALARBEIT required for autonomy, robustness, and safety in the system. (iii) Fi- studied by Kaindl et al. [11]. An agent is defined as a combination nally, the third challenge regards the communication and decentral- of hardware and software components which represents itself and ized existence of information in a distributed system, operating in its relations to its environment in an explicit symbolic manner. The known or unknown environments. proposed system is used for self-configuration as well as monitoring In this paper, we propose an architecture designed to address and failure detection. However, no cause detection was considered in that work. these individual challenges in an integrated and comprehensive Self-monitoring has been implemented also as a hierarchical manner in order to meet the expectations of modern production agent in a Systems-on-Chip (SoC) [12]. The proposed structure fa- systems. Therefore, the proposed system aims at the dynamic ob- cilitates the process of monitoring parallel many-core SoCs. Cyber- servation of the environment, data abstraction, and distributed ne- Physical System-on-Chip is another platform where self-awareness gotiation. They serve each unit of the system in detecting the faults has been explored [13]. There, the authors described self-awareness and anomalies independently but collaboratively. Each unit uses self- as the ability of a system to monitor its own internal and external be- configuration to achieve dynamic clustering of the system for the havior in order to make appropriate decisions. The superiority of the purpose of communication between units. Finally, the system at- hereby proposed architecture of self-aware health monitoring over tempts to go beyond using the existing actions for known prob- the state of the art is its ability to interpret the reliability of the data, lems and to mitigate new anomalies and problems to guarantee the measure the confidence of its processes, use attention for more ef- safety of the whole system. ficient resource utilization, and perform predictions to provide more It should be born in mind that the proposed architecture is work information for the decision-making process. in progress and not yet completely implemented. We have sim- ulated the SAMBA architecture in the nxtStudio based environ- 2.2 Cognitive systems ment and verified its main functionality, which we present here. Flexible and reliable operation in many automated processes is We have implemented a portion of Self-Aware Health Monitoring achieved by the inclusion of the human operator in the loop [14]. (SAHM), namely health monitoring of an injective function black Cognitive abilities enable humans to solve problems under uncer- box using contextual information, which we tested on an AC- tainty and despite the changes in the environment and tasks. Al- motor case study [8]. Moreover, we have recently studied the util- though such capabilities are common in humans, they are rarely ity of SAMBA features in the context of hierarchical goal manage- found in industrial systems. The goal of cognitive architectures is ment and found encouraging preliminary results for their use in to define a framework for designing systems with human-like intel- autonomous anomaly mitigation, self-configuration and arbitration ligence; they provide a structure which enables a system to develop between conflicting goals [9]. Although we are encouraged by the over its lifetime by embedding the mechanisms of perception, rea- results so far, the main challenge will be to demonstrate that SAMBA soning, action, and learning [15]. Different cognitive architectures leads to a measurable improvement in overall reliability and opera- have been applied for realizing cognitive tasks, for example, SOAR tional efficiency of the production system. That is the target of an [16], LIDA [17] and ACT-R [18]. ongoing FFG funded project with TU Wien, Danube University Krems To address limitations related to incomplete sensing of the envi- (DUK), nxtControl, and AVL (Anstalt für Verbrennungskraftmaschi- ronment and inability to individually carry-out global tasks, cogni- nen List) as partners. tive systems often exhibit social behavior, which entails communica- This paper first reviews the state of the art in the areas of tion and cooperation or negotiation. Such an approach is adopted self-aware monitoring, cognitive systems and dynamic clustering in in cognitive radio networks, where multiple distributed sensors col- Sect. 2. In Sect. 3, it explains how these methods are integrated into laborate selectively to enhance their spectrum sensing performance the SAMBA architecture and presents the modules of the solution and the utilization of the radio frequency spectrum [19]. Distributed and their interfaces. Next, in Sect. 4, it goes through an exemplary cognitive systems have been studied also for the control of robots scenario with disturbances to elaborate how the system handles the [20], and building environmental control [21]. In industrial appli- problem and presents a high-level simulation of the system using cations, the cognitive production systems refer to highly intercon- IEC 61499 function blocks. Finally, it discusses potential benefits of nected devices with improved sensing, reasoning, learning and plan- the proposed architecture and draws conclusions. ning capabilities, which use knowledge-based and learning mod- els to assess and expand their capabilities [1]. The cognitive system 2. State of the art design paradigm provides a promising approach towards the dy- namic adaptation of the processes and the continuous optimization 2.1 Self-aware monitoring through learning by observation of the environment and reasoning Recent attempts to improve the efficiency, collaboration and re- for mitigating errors and finding improved operation strategies. silience of automated systems in industry highlight the importance A more recent cognitive architecture is the Simulation of Mental Apparatus (SiMA) [22], which has previously been applied for cog- of the self-awareness in such systems. Self-awareness enables a sys- nitive control in building automation [23]. SiMA focuses on func- tem to monitor itself and its own environment to assess its situation tions that generate human behavior and implements the underly- better and make more appropriate decisions. An architecture for ing mechanisms (e.g., drives, emotions) which drive a system to cyber-physical manufacturing system based on Industry 4.0 has been exhibit a certain behavior. This behavior is thereby defined by the proposed by Lee et al. [10]. The authors studied a unified 5-level ar- internal state of the system, which on its turn depends on the feed- chitecture. The first level deals with the data acquisition and then back returned by the environment on the system’s actions, rather the self-awareness is used in the second level to monitor the health than being explicitly defined by the environmental state. Although degradation of the system. In another study, self-representation for this approach requires a higher engineering effort in the beginning monitoring automation systems using a multi-agent structure was due to the need for implementing the underlying driving behavioral mechanism, it can achieve a lower system complexity since the states under which a behavior is exposed do not need to be defined ex- SAVE (Self-monitoring-based process Adaptation for quality assurance in het- erogeneous VErsatile manufacturing), funded by FFG under contract 864883. plicitly [24]. The SiMA theory is used as basis for the design of the Juni 2018 135. Jahrgang The Author(s) heft 3.2018 271 ORIGINALARBEIT L. C. Siafara et al. SAMBA – an architecture for adaptive cognitive control... cognitive module in SAMBA, which is discussed in detail in the next section. 2.3 Dynamic clustering A key to an efficient performance in distributed systems is commu- nication among system components. The concept of clustering was devised in conjunction with the need to adapt to high complex- ity and the dynamic nature of emerging manufacturing control sys- tems. It originated from the Holonic Manufacturing Systems (HMS) concept [25] and the MetaMorph approach of HMS [26], based on the idea of expanding problem-solving clusters. Each cluster is a dy- namically created community of intelligent system components that cooperate with each other to obtain enough information to solve a problem. Each member of a cluster has a set of algorithms which enable recognition of a problem and finding a solution. A flexible structure is one of the advantages of a distributed clustering con- cept that allows solving complex problems in a scalable manner. Clustering has found its use in a variety of industrial applications such as mitigating the complexity of production orders [27], knowl- Fig. 1. The architecture of SAMBA depicting the structure of one En- edge propagation of thermal modeling [28], ad-hoc wireless net- tity works [29] for optimizing the routing protocol [30], andinother concepts based on swarm intelligence [31, 32]. The proposed architecture uses the concept of distributed clus- 3.2 Self-aware knowledge extraction and representation tering for establishing relevant connections between independently SAHM aims at parallel data abstraction and health status analysis acting entities in the system. Similar to [27], it establishes connec- (anomaly detection) to provide the system with a suitable observa- tions between the system components according to the production tion required for self-awareness [33]. The data is abstracted in a order structure. In this case, however, the goal (rather than decision- normal situation when the sensors’ functionality is correct. How- making) is more relevant to knowledge propagation [28] and there- ever, when some of the sensors are faulty, the intention is to request fore, the architecture presents a novelty over the current state of the equivalent data from adjacent entities. The health status analysis the art. approach mostly relies on the self-observation, however, sometimes negotiation with other entities increases the knowledge of an entity 3. Proposed architecture regarding an anomaly. Therefore, the proposed self-aware system Figure 1 illustrates the architecture of SAMBA and its major mod- monitors various health parameters in a distributed manner. These ules. The logical unit of an entity is an Autonomous Cooperating parameters are used to classify the health status of the entity into a Object (ACO). The ACO learns locally the specifics of the environ- predefined state. The input parameters are the data of the sensors ment and the actions it can take. Further, it exhibits social behavior, that have been abstracted. These sensors may belong to an entity since it interacts with other ACOs located in the same environment. itself or to other entities. To decide on its agenda, it takes into consideration its own goals, The goal of abstraction is to reduce the size of data as well as the environmental context and the requests from other ACOs. The to extract knowledge which is beneficial for the further processes global behavior emerges out of the interactions among the ACOs. in SAHM, the internal manager, or other entities. It deals with the sensor data and also other specifications about the hardware part 3.1 System architecture of the entity, including the nominal range and the reliability of each Each ACO consists of three SAMBA-specific functional units in addi- component. Abstraction has several functional states, e.g., abstrac- tion to the operation module: Operation Module serves as the inter- tion of sensor data, providing a response to the internal manager’s face to the hardware. It connects to the plant controller and forms a request for specific data as well as providing specific data to the major part of the by providing symbolized sensor data and receiving external manager, or processing the data provided from another ex- actuator ACO commands. Often, this unit sits on the legacy con- ternal manager. troller. SAHM monitors the system operation and in case of anoma- The goal of health status analysis is to detect the anomaly (as lies, it proposes possible causes. To this end, SAHM performs fault well as its characteristics, causes, and effects), and to inform the diagnosis and data abstraction. As each ACO is observing only its internal manager. It also deals with the requests from the external local environment, other ACOs can request or may be requested manager regarding the health and operational status of the entity. Health status analysis includes anomaly detection, anomaly specifi- to provide further information through the external manager. Inter- cation, cause diagnosis and prediction blocks. The abstracted data nal Manager is the decision-making entity of the system. It receives first enters the anomaly detection. The type of anomaly and the health status information from SAHM and external requests from value of the reliability are the outputs which are forwarded to the other ACOs through the external manager. At a certain state, there other three blocks. The detected anomaly triggers the anomaly spec- might be conflicting action requests and the internal manager shall ification block which estimates the features of this anomaly, such decide which goal is prioritized. External Manager provides an im- as the rate of wear-out deterioration. Cause diagnosis detects the plicit connectivity among the relevant ACOs via distributed cluster- reason(s) of the anomaly. This block requires the abstracted data, ing based on the production order structure and the physical layout type of anomaly and features of the anomaly (if they exist) as well of the shop floor. It facilitates transparent negotiations between one as complementary information from other ACOs. If the health status ACO and other ACOs in the cluster. In the rest of this section, we analysis suspects the existence of an external cause, a data request is discuss the main functionalities of each module. 272 heft 3.2018 The Author(s) e&i elektrotechnik und informationstechnik L. C. Siafara et al. SAMBA – an architecture for adaptive cognitive control... ORIGINALARBEIT sent to the external manager, which can ask for the data and health have not been activated for a long time. Decisions of the Internal status of the other involved ACOs. Another output of this block is Manager (IM) are evaluated and their effects are updated over time the new value of specification for the reliability of the components in order to account for the changing dynamics of the environment, which is transferred to the abstraction block. Finally, the prediction therefore, maintaining an implicit model of the environment. block receives the abstracted data, type of anomaly, its features, and cause(s), and predicts the future effect of the anomaly on the entity 3.4 Adaptive collaboration or the system. To compensate for the lack of a global overview due to the dis- The frequently updated reliability information of the data and tributed nature of the proposed system, the concept of dynamic components provided from MES as well as the confidence value clustering is applied. Dynamic clustering introduces a flexible way at the end of each step are all interpreted as metrics to measure to integrate global objectives while allowing encapsulation of core the level of uncertainty. Different approaches are used depending functionality of the ACO, hence, providing transparency for the on where the uncertainty is flagged. For example, if the abstrac- decision-making. The challenge is to form clusters regarding the tion unit faces unreliable data, a negotiation with the neighboring functional requirements of the shop floor and manufacturing in- ACO(s) starts to request data from them. A self-aware Multiple Clas- structions. Currently, there is no methodology to formalize produc- sifier System (MCS) based on the rankings of each individual clas- tion order ontology in relation to the anomalies. Instead of pre- sifier similar to [34] can be used to improve the performance of defined rules of the existing solutions, the External Manager (EM) anomaly detection and cause diagnosis through analysis of a col- defines dynamic methods using learning algorithms (e.g., neural lecting of results and thereby choosing the optimal one to reduce networks, and reinforcement learning) to automatically discover the the uncertainty. Cause detection can be built using a MCS whose cluster formation rules based on the production order, the physical inputs are the type of anomaly, meta-data and anomaly parameters. layout of the shop floor, and network communication structure. A Multiple learning algorithms interpret the inputs and provide the re- set of generic classifiers which operate on non-application-specific sult with a value of confidence. If an external cause is probable to characteristics are defined and used by the EM to determine es- obtain relevant external data, negotiation with other ACOs is initial- sential connections in the cluster and establish weighted links be- ized. The new external information is fed as an input to the system tween ACOs based on the prioritized set of control parameters. and a new analysis starts. These weighted connections are updated at runtime using back- propagation according to feedback from the evaluation of the com- 3.3 Cognitive decision-making munication and the execution of decisions. To select the most suitable actions with regard to the goals of an As a result, the system achieves implicit connectivity among the ACO, decision-making takes place on the basis of data inputs from relevant ACOs and facilitates transparent negotiations between the SAHM and the external manager. SAHM provides information SAHM and the internal manager of different ACOs in a cluster. The regarding the current state of the ACO and its environment, whereas External Manager bases its technique on a non-deterministic rele- the external manager provides information regarding the state of vance of connection between individual ACOs regarding a particular other ACOs and their environment. Two input types are defined: event. The outputs of the External Manager are used to check possi- drives and percepts. Drives are the (initial) motivations of the system ble implications of local decisions on other ACOs in the community. and therefore, the source of the system goals. Their intensity rep- By adjusting the relevance of the connections over time, such de- resents the deviation from a desired goal state: the higher a drive pendencies are dynamically adapted based on the consequences of intensity, the more urgent the goal represented by this drive. Sym- actions. bols from the environment constitute the percepts of the ACO. Once percepts are generated for the current state, similar states from the 3.5 System dynamics and interface memory are activated. Emotions are the internal evaluation mech- Communication between the units of an ACO and also between dif- anism of the system and are generated based on the current state ferent ACOs of a cluster is necessary to achieve the functionality of of drives and the stored activated memories from the past. All these the larger system described in Fig. 2. This figure highlights the wiring processes form part of the primary process of the system, which is and content of data exchanges between units in Fig. 1. The environ- characterized by the lack of reasoning functions. The primary pro- ment consists of the yellow blocks on the top, exchanging informa- cess regulates the reactive behavior of the system and proposes tion with the Operation Module in the ACO, and other ACOs shown actions that are able to solve urgent problems in a reactive man- at the bottom. Types of information transferred between units inside ner [35]. the ACO are also specified and depicted in Fig. 2. Finally, the ACO In the secondary process, social rules which reward or penal- is connected to the External Manager of other ACOs, which form ize a behavior are tested on the current state; they represent user the other part of the environment. Note that this figure illustrates preferences and guide the policies that the system should comply the minimum possible connections in the proposed system. Each re- with them. The rewards, which represent the external evaluations, quest and its reply are labeled by an ID to simplify the process of along with the emotions, which are the internal evaluations, en- handling them. To focus on the main system functionality, it is as- rich the current perceived state. The enriched perceived state and sumed that no communication errors occur. the drives, which indicate goal priorities, are the inputs to the goal The connection between the Internal Manager and the Operation selection process, where it is decided which goal the system shall Module is established for transferring the estimated set-points after pursue. Next, sequences of states (episodes) similar to the current the decision-making process. There is only a one-directional data episode are activated using case-based reasoning. From the acti- flow from the Operation Module to SAHM to push all data from vated episodes, sequences of actions (policies) that managed to re- sensors to SAHM. duce the drive intensities in the past are then picked as potential Several communication messages are separately sent from SAHM options. The policy with the best performance is used to select the current action to execute. The new episode is saved in the memory to the Internal Manager including a value representing the health by the learning process, which is responsible for the tasks of updat- status of the ACO, information on anomalies and their cause, ex- ing the memory with new episodes while it removes episodes that tra data in response to Internal Manager’s requests, and the action Juni 2018 135. Jahrgang The Author(s) heft 3.2018 273 ORIGINALARBEIT L. C. Siafara et al. SAMBA – an architecture for adaptive cognitive control... conveyor sections supplying two robotic arms with objects. includes three different types of entities, i.e., the conveyor sections, which transport the products, the product sorters, which sort the products by type for further processing, and the robotic arms, which process the products. Each entity consists of the hardware part (i.e., mo- tor, sensors, and other equipment), and the logic unit (the ACO), which monitors and, if necessary, adjusts its operation. The system monitors the existence of all items in its environment. If an ACO is expecting to receive an item at a specific time but it does not, the system shall try to find the root cause of the event. To this end, it uses the observations from entities responsible for previous (or next) tasks in the process and analyzes the available information to find out if the item is missing or the sensor is not working. Afterwards, the system needs to decide whether it can adjust the settings and resolve the problem, or the operator needs to be notified. The following tasks take place: 1. The SAHM detects the event and tries to diagnose the problem. That is, to find out whether the problem is caused by a faulty sen- sor or a missing product. It informs the internal manager about the event. 2. The external manager, upon notification from the internal man- ager, asks for sensor information from the previous/next ACOs in the process. 3. The SAHM receives the information from the other related ACOs, and analyzes all information to find out if the item is missing or the sensor is not working. This action will continue in a sequence until the source of the problem (product missing/stuck or sensor Fig. 2. Communication flow within the ACO and with other ACOs of failure) is found. the cluster 4. If the problem is due to erroneous sensor data, the SAHM notifies the internal manager for this event to take the necessary actions to handle the problem. proposal. The action proposal responds to a request made by the 5. If the problem is that the product is missing, due to unsynchro- Internal Manager and indicates the operational range within which nized speed settings of the source entities, the SAHM notifies the ACO can adapt its operational settings. Data and action requests the internal manager and the latter takes the necessary actions are sent from the Internal Manager to SAHM. Action requests are to handle the problem. sent to SAHM upon receipt of an action request from the external 6. If the problem is that the product is lost, then the internal man- manager, in order to ask for the nominal operational range of the ager informs the human operator about the fault and asks for ACO. intervention. The Internal Manager establishes a connection to the External Manager when the decision it makes impacts the operational pa- rameters of another ACO. An action request is also sent when the 4.2 Feasibility verification of communication architecture Internal Manager decides to change an operation parameter and the To verify the feasibility of the communication architecture, we ran negotiation needs to be initiated. Moreover, in response to an estab- a simulation experiment which we present in the rest of this sec- lished negotiation from other ACOs, the Internal Manager sends a tion. The simulation was performed in the nxtStudio runtime envi- confirmation message to the requesting ACO in case of an agree- ronment which uses the IEC 61499 standard. The standard is based ment. on the event-driven concept of function blocks. Function blocks are The messages from SAHM to the External Manager include data independently acting components that encapsulate local function- requests when SAHM detects an anomaly and needs to contact ality and communicate with each other via events and associated another ACO for additional information. Similar messages are ex- data. Each block implements a set of algorithms and triggers events changed in the opposite direction too. The communication mecha- upon completion. The main benefit of the IEC 61499 model lies in nisms between External Managers of different ACOs are similar to its flexibility to integrate complex systems and adapt them even at the ones between the External Manager and the Internal Manager. runtime. In addition, External Managers exchange data requests to establish The main goal of the simulation was to establish a communication clusters. infrastructure for testing the synergy of the individual SAMBA com- ponents. For each node in the production line of the use case (Fig. 3), 4. Exemplary use case a corresponding ACO was created. The general structure of an ACO is showninFig. 4; the ACO is represented as a composite function 4.1 Definition/description block with SAHM, IM, EM and OM as basic function blocks. Event To explain how the system behavior emerges out of the interactions inputs and outputs with associated data were defined and the com- among the individual components of the ACOs as well as through munication algorithms were implemented. The resulting simulation communication with other ACOs, consider an exemplary scenario with detection of a missing product. Figure 3 illustrates a poten- provides a flexible environment for testing the underlying algorithms tial configuration of the system; an assembly line consisting of four of SAMBA that can be used as an application independent runtime 274 heft 3.2018 The Author(s) e&i elektrotechnik und informationstechnik L. C. Siafara et al. SAMBA – an architecture for adaptive cognitive control... ORIGINALARBEIT Fig. 3. Shop floor configuration in the missing product use case Fig. 4. ACO function-block structure environment. Due to the modular nature of the IEC 6149 standard, The design of SAMBA addresses following challenges which mod- a variety of use cases and algorithms can be integrated and tested ern industrial production process faces: to analyze the optimal configuration of algorithms and parameters 1. Increased adaptivity and reduced engineering effort: to interpret for a particular application. The simulation results showed the sat- the context of operation with minimum a priori knowledge avail- isfying level of scalability and flexibility of the proposed architecture able, semantic enhancement of the available information is used. and its ability to integrate required algorithms for data acquisition This is enabled by self- and context-awareness concepts such as and decision-making. confidence, attention, data-reliability, and history, combined with intelligent data analysis, such as data abstraction and scattered- 5. Discussion data fusion. SAMBA is an architecture for CPPS that monitors the production 2. Quality assurance and resilience enhancement through fault process and reacts to deviations, either through automatic compen- diagnosis and prognosis: in a distributed system no single sations or by informing the operator. It is designed to operate as a (sub)system knows the complete state of the overall system. middleware on the top of existing (legacy) systems adding a layer Communication with other ACOs takes place to enhance local of intelligence to the CPPS. Intelligence in this specific context is information, and to increase confidence about local knowledge. defined as the ability of the system to be context-aware and self- This also helps the system to obtain awareness of the bigger aware, to be proactive and social, acting within a certain degree of context in order to detect, analyze, predict, and mitigate errors, autonomy in a changing environment. To this end, SAMBA builds faults, and failures. upon the concept of the ACO, as presented above. The behavior of 3. Enhanced autonomy and intelligence for mitigation of anomalies the large system emerges out of the cooperation of individual ACOs, in operation: in complex manufacturing systems (such as CPPSs) which dynamically form clusters and collaborate on demand while thorough and precise modeling of the system and its environ- pursuing their own goals as well as the goals of the global system. To ment is challenging. Cognitive systems are adept at making deci- compensate for the lack of central supervision, on one hand, ACOs communicate with each other to enhance their knowledge and un- sions efficiently despite the lack of a complete or precise model. derstand the context of their observations. On the other hand, they Cognitive decision-making improves the efficiency of the system negotiate to align their actions in order to efficiently achieve the in using extracted knowledge about each ACO, its neighbors in global system goals. the cluster, the overall system, and possible courses of action, to Juni 2018 135. Jahrgang The Author(s) heft 3.2018 275 ORIGINALARBEIT L. C. Siafara et al. SAMBA – an architecture for adaptive cognitive control... re- or pro-actively change the CPPS, regarding the timing con- 5. Zhang, Y., Qian, C., Lv, J., Liu, Y. (2017): Agent and cyber-physical system based self- organizing and self-adaptive intelligent shopfloor. IEEE Trans. Ind. Inform., 13(2), 737– straints at hand. Especially by mitigation of failures, degraded health, or anomalies, it ensures the quality of the product and 6. Park, H.-S., Tran, N.-H. (2012): An autonomous manufacturing system based on swarm process in adaptive heterogeneous manufacturing systems. of cognitive agents. J. Manuf. Syst., 31(3), 337–348. 4. Dynamic clustering based on environment discovery: dynamic 7. Monostori, L. (2014): Cyber-physical production systems: roots, expectations and r&d cluster building concepts which take into account the produc- challenges. Proc. CIRP, 17, 9–13. 8. Götzinger, M., TaheriNejad, N., Kholerdi, H. A., Jantsch, A. (2017): On the design of tion orders parsing allow effective negotiation and propagation context-aware health monitoring without a priori knowledge; an ac-motor case-study. of mitigation measures. They also enable representation of the In 2017 IEEE 30th Canadian conf. on electrical and comp. eng. (CCECE) (pp. 1–5). overall production environment beyond its locally accessible in- 9. Jantsch, A., Anzanpour, A., Kolerdi, H., Azimi, I., Siafara, L. C., Rahmani, A. M., Taher- formation. inejad, N., Liljeberg, P., Dutt, N. (2018): Hierarchical dynamic goal management for IoT systems. In Proc. of the IEEE int. symp. on quality electronic design, USA. SAMBA aims to introduce a new design paradigm in the industrial 10. Lee, J., Bagheri, B., Kao, H.-A. (2015): A cyber-physical systems architecture for indus- try 4.0-based manufacturing systems. Manufac. Lett., 3, 18–23. environment which, thanks to the generic nature of the proposed 11. Kaindl, H., Vallée, M., Arnautovic, E. (2013): Self-representation for self-configuration architecture, allows integration of legacy systems in manufacturing and monitoring in agent-based flexible automation systems. IEEE Trans. Syst. Man with considerably less effort compared to the state of the art. The Cybern., 43(1), 164–175. generic nature of the architecture makes it possible to apply it inde- 12. Guang, L., Plosila, J., Isoaho, J., Tenhunen, H. (2010): Hierarchical agent monitored pendent of the specifics of the entities it controls. It is expected to parallel on-chip system: a novel design paradigm and its formal specification. Int. J. Embed. Real-Time Commun. Syst., 1(2), 85–105. reduce the downtime and frequency of maintenance interventions 13. Sarma, S., Dutt, N., Gupta, P., Nicolau, A., Venkatasubramanian, N. (2014): On-chip by health monitoring, predictive analysis and mitigation of behav- self-awareness using cyberphysical-systems-on-chip (CPSOC). In Proc. intl. conf. on ioral deviations. Next steps in the development of SAMBA include hardware/software codesign and system synthesis (p. 22). New York: ACM. the detailed design of the algorithms for the modules of the ar- 14. Zhao, Y. F., Xu, X. (2010): Enabling cognitive manufacturing through automated on- chitecture and testing of the emergent system performance. In the machine measurement planning and feedback. Adv. Eng. Inform., 24(3), 269–284. 15. Vernon, D. (2017): Two ways (not) to design a cognitive architecture. Cognitive Robot planned follow-up project, SAVE, the challenge will be to demon- Architectures, 42. strate – quantitatively – that SAMBA improves the overall system 16. Laird, J. E. (2012): The Soar cognitive architecture. Cambridge: MIT Press. performance in terms of reliability and operation efficiency. 17. Snaider, J., McCall, R., Franklin, S. (2011): The lida framework as a general tool for agi. In Int. conf. on AGI. Berlin: Springer. 18. Taatgen, N. A., Lebiere, C., Anderson, J. R. (2006): Modeling paradigms in act-r. Cog- 6. Conclusion nition and multi-agent interaction: from cognitive modeling to social simulation, pp. In this paper, we presented SAMBA, an architecture for versatile 29–52. heterogeneous CPPS with focus on predictive analysis, autonomous 19. Yucek, T., Arslan, H. (2009): A survey of spectrum sensing algorithms for cognitive health monitoring, and error mitigation in industrial manufacturing radio applications. IEEE Commun. Surv. Tutor., 11(1), 116–130. processes. The synergies provided by these functionalities reduce 20. Kawamura, K., Peters, R. A. II., Bodenheimer, R. E., Sarkar, N., Park, J., Clifton, C. A., Spratley, A. W., Hambuchen, K. A. (2004): A parallel distributed cognitive control system reaction time and engineering efforts through adjustment system for a humanoid robot. Int. J. Humanoid Robot., 1(01), 65–93. of production steps to compensate for the deviations. SAMBA in- 21. Kollmann, S., Siafara, L. C., Schaat, S., Wendt, A. (2016): Towards a cognitive multi- creases the ability of the system to assure the quality of the product agent system for building control. Proc. Comput. Sci., 88, 191–197. and the process by intelligently adapting to rapidly changing envi- 22. Schaat, S., Wendt, A., Kollmann, S., Gelbard, F., Jakubec, M. (2015): Interdisciplinary development and evaluation of cognitive architectures exemplified with the SIMA ap- ronments. The architecture is mainly developed for distributed CPPS, proach. In EAPCogSci. however, its generic and modular design enables its future applica- 23. Zucker, G., Habib, U., Blöchle, M., Wendt, A., Schaat, S., Siafara, L. C. (2015): Building tion to different industrial systems with minimum effort for adapta- energy management and data analytics. In EDST, 2015 int. symposium (pp. 462–467). tion. 24. Wendt, A., Kollmann, S., Siafara, L., Biletskiy, Y. (2018): Usage of cognitive architec- tures in the development of industrial applications. In ICAART. 25. Van Brussel, H., Wyns, J., Valckenaers, P., Bongaerts, L., Peeters, P. (1998): Reference Acknowledgements architecture for holonic manufacturing systems: Prosa. Comput. Ind., 37(3), 255–274. Open access funding provided by TU Wien (TUW). We gratefully 26. Kühnle, H. (2009): Distributed manufacturing: paradigm, concepts, solutions and ex- acknowledge the financial support provided to us by the BMVIT/FFG amples. Berlin: Springer. under the program ICT of the Future in the project SAMBA (contract 27. Babiceanu, R. F., Chen, F. F. (2006): Development and applications of holonic manu- facturing systems: a survey. J. Intell. Manuf., 17(1), 111–131. number: 855426). 28. Bratukhin, A., Nagy, A., Mahmood, A. (2015): Distribution of control functionality in energy-aware industrial building environment. In WFCS, IEEE world conf. (pp. 1–8). Open Access This article is distributed under the terms of the Creative New York: IEEE Press. Commons Attribution 4.0 International License (http://creativecommons.org/ 29. Abbasi, A., Younis, M. (2007): A survey on clustering algorithms for wireless sensor licenses/by/4.0/), which permits unrestricted use, distribution, and reproduc- networks. Comput. Commun., 30(14), 2826–2841. tion in any medium, provided you give appropriate credit to the original au- 30. Liu, X. (2012): A survey on clustering routing protocols in wireless sensor networks. thor(s) and the source, provide a link to the Creative Commons license, and Sensors, 12(8), 11113–11153. indicate if changes were made. 31. Saleem, M., Di Caro, G. A., Farooq, M. (2011): Swarm intelligence based routing protocol for wireless sensor networks: survey and future directions. Inf. Sci., 181(20), References 4597–4624. 32. Kennedy, J. (2006): Handbook of nature-inspired and innovative computing (pp. 187– 1. Zaeh, M. F., Reinhart, G., Ostgathe, M., Geiger, F., Lau, C. (2010): A holistic approach 219). Berlin: Springer. for the cognitive control of production systems. Adv. Eng. Inform., 24(3), 300–307. 33. Taherinejad, N., Jantsch, A., Pollreisz, D. (2016): Comprehensive observation and its 2. Shea, K. (2010): The cognitive factory. role in self-awareness – an emotion recognition system example. In Conf. on comp. 3. European Factories of the Future Research Ass. et al. (2013): Factories of the future: science and inf. sys. Multi-annual roadmap for the contractual ppp under horizon 2020. Pub. office of the 34. Kholerdi, H. A., TaheriNejad, N., Jantsch, A. (2018): Enhancement of classification of Eur. Union: Brussels. small data sets using self-awareness – an iris flower case-study. In Proc. of the int. 4. Armengaud, E., Sams, C., von Falck, G., List, G., Kreiner, C., Riel, A. (2017): Industry symp. on circuit and systems (ISCAS), Florence, Italy. 4.0 as digitalization over the entire product lifecycle: opportunities in the automo- 35. Zucker, G., Wendt, A., Siafara, L., Schaat, S. (2016): A cognitive architecture for build- tive domain. In Eur. conf. on software process improvement (pp. 334–351). Berlin: ing automation. In Ind. electronics society, IECON 2016 (pp. 6919–6924). New York: Springer. IEEE Press. 276 heft 3.2018 The Author(s) e&i elektrotechnik und informationstechnik L. C. Siafara et al. SAMBA – an architecture for adaptive cognitive control... ORIGINALARBEIT Authors current research focus is in the area of distributed control sys- Lydia Chaido Siafara tems, artificial intelligence, machine learning and modeling in in- received her degree in electrical engineer- dustrial and building automation with the focus on efficient energy ing and computer technology from Aristo- use. tle University of Thessaloniki, Greece, and the master’s degree in building science and technologies from Vienna University of Tech- Nima Taherinejad nology (TU Wien) in 2014. Since 2015 she is a Ph.D. graduate of The University of British has been working as a project assistant at Columbia, Vancouver, BC, Canada. He is cur- TU Wien, where she was responsible for the rently a Post-doctorate University Assistant at KORE project, dealing with the application of TU Wien, Vienna, Austria, where his main ar- human-inspired cognitive systems in building automation. Her work eas of research include systems on chip, self- is concerned with the development of intelligent agents and their awareness in cyber-physical systems, embed- application in the field of automation systems. Her research interests ded systems, and robotics. He was the gen- include artificial intelligence, cognitive systems and machine learn- eral chair and TPC chair of MobiHealth 2017, ing. chair of the AUTO21 HQP Advisory Commit- tee and a member of board of directors of AUTO21 Center of Excel- Hedyeh Kholerdi lence. He has authored a book and published in/served as a reviewer received her master’s degree in the field of for various journals and conferences. Dr. Taherinejad has received Image Processing from Babol University of several awards and scholarships from universities and conferences Technology, Iran. Her master’s thesis is titled he has attended. as “Drowsiness detection using image pro- cessing technique inspired by the human vi- Axel Jantsch sual system”. She joined TU Wien, Institute received the Dipl.-Ing. and Dr. techn. degrees of Computer Technology (ICT) as a university from TU Wien, Vienna, Austria, in 1988 and assistant for Embedded Systems and Online 1992, respectively. From 1997 to 2002, he VHDL Programming courses. She was also a was an Associate Professor with the KTH project assistant working on self-aware health monitoring in scalable Royal Institute of Technology, Stockholm, distributed systems. The main focus was on research in requirement Sweden, where he was also a Full Professor of creating a contextual awareness in a distributed autonomous sys- of Electronic Systems Design from 2002 to tem using learning approaches. 2014. Since 2014, he has been a Professor with the Institute of Computer Technology, Aleksey Bratukhin TU Wien. He has authored over 300 articles and one book in the received his master’s degree at Perm Techni- areas of VLSI design and synthesis, HW/SW codesign and cosyn- cal University in Russia in 1998. From 2000 thesis, networks-on-chip, and self-awareness in Cyber-Physical Sys- to 2006 he worked at Vienna University of tems. Technology with the focus on software agent systems and distributed control in respect to vertical integration in the area of plant au- tomation, where he received his doctorate degree in 2006. In that year he joined the Center for Integrated Sensor Systems. His Juni 2018 135. Jahrgang The Author(s) heft 3.2018 277 http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png e & i Elektrotechnik und Informationstechnik Springer Journals

SAMBA – an architecture for adaptive cognitive control of distributed Cyber-Physical Production Systems based on its self-awareness

Free
8 pages
Loading next page...
 
/lp/springer_journal/samba-an-architecture-for-adaptive-cognitive-control-of-distributed-2DFUwn6Sn0
Publisher
Springer Vienna
Copyright
Copyright © 2018 by The Author(s)
Subject
Engineering; Electrical Engineering; Computer Hardware; Software Engineering/Programming and Operating Systems
ISSN
0932-383X
eISSN
1613-7620
D.O.I.
10.1007/s00502-018-0614-7
Publisher site
See Article on Publisher Site

Abstract

ORIGINALARBEIT Elektrotechnik & Informationstechnik (2018) 135/3: 270–277. https://doi.org/10.1007/s00502-018-0614-7 SAMBA – an architecture for adaptive cognitive control of distributed Cyber-Physical Production Systems based on its self-awareness L. C. Siafara, H. Kholerdi, A. Bratukhin, N. Taherinejad, A. Jantsch Factories in Industry 4.0 are growing in complexity due to the incorporation of a large number of Cyber-Physical System (CPSs) which are logically and often physically distributed. Traditional monolithic control and monitoring structures are not able to address the increasing requirements regarding flexibility, operational time, and efficiency as well as resilience. Self-Aware health Monitoring and Bio-inspired coordination for distributed Automation systems (SAMBA) is a cognitive application architecture which processes information from the factory floor and interacts with the Manufacturing Execution System (MES) to enable automated control and supervision of decentralized CPSs. The proposed architecture increases the ability of the system to ensure the quality of the process by intelligently adapting to rapidly changing environments and conditions. Keywords: self-awareness; cognitive systems; dynamic clustering; autonomous collaborating objects; system health monitoring SAMBA – eine Architektur zur adaptiven kognitiven Kontrolle verteilter cyber-physischer Produktionssysteme basierend auf Self-Awareness. Industrie 4.0-Fabriken nehmen rasch an Komplexität zu aufgrund der Einbeziehung einer großen Anzahl von cyber-physischer Syste- me (CPS), die logisch und oft physisch verteilt sind. Traditionelle monolithische Kontrolle und Überwachungsstrukturen sind nicht in der Lage, den steigenden Anforderungen hinsichtlich Flexibilität, Betriebszeit und Effizienz sowie auch Belastbarkeit gerecht zu wer- den. „Self-Aware Health Monitoring and Bio-inspired coordination for distributed Automation systems“ (SAMBA) ist eine kognitive Anwendungsarchitektur, die Informationen von der Fabrik verarbeitet und mit dem Manufacturing Execution System (MES) zur auto- matisierten Kontrolle und Überwachung von dezentralen CPS interagiert. Die vorgeschlagene Architektur erhöht die Fähigkeit eines Systems, durch intelligente Anpassung an eine sich schnell verändernde Umgebung bzw. Bedingungen die Qualität des Prozesses zu gewährleisten. Schlüsselwörter: Self-Awareness; kognitive Systeme; dynamisches Clustering; Autonomous Collaborating Objects; System Health Monitoring Received January 30, 2018, accepted April 14, 2018, published online June 4, 2018 © The Author(s) 2018 1. Introduction optimization and a number of sophisticated cognitive architectures Highly efficient production requires high degrees of flexibility, adap- [5, 6]. Even though the developments of technology have improved tiveness, and responsiveness in order to achieve high quality and the robustness and resiliency of Cyber-Physical Production System versatility of manufacturing processes [1]. To reduce lead time, previ- (CPPSs) in comparison to conventional automation systems, new ous approaches have focused on rigid and deterministic automated expectations such as self-diagnosis and prognosis, self-repair, self- production environments, which minimize disturbances during op- discovery and self-configuration, predictability as well as safety [7] eration [2]. However, the increasing structural complexity of produc- are rising. We categorize these expectations into three different tion systems, due to the growing number of CPSs and distributed challenges. (i) The first challenge is the self-aware health monitoring heterogeneous components in the loop, decrease the determinis- which means self-observation and fault diagnosis. (ii) The second tic nature of production processes and require agile controls capa- challenge regards the complication of the decision-making process ble of prediction and timely reaction to disturbances [3]. To achieve flexibility while enhancing system performance, real-time informa- tion from the shop floor shall be integrated into the control system. Siafara, Lydia Chaido, TU Wien, Institute of Computer Technology, Gußhausstraße Moreover, the optimal reaction should be decided with respect to 27–29, 1040 Vienna, Austria; Kholerdi, Hedyeh, TU Wien, Institute of Computer Technology, Gußhausstraße 27–29, 1040 Vienna, Austria; Bratukhin, Aleksey, Danube the system goals, which may themselves change during the oper- University Krems, Center for Integrated Sensor Systems, Viktor Kaplan Straße 2 E, 2700 ation. To further assure quality, an important task is the continu- Wiener Neustadt, Austria; Taherinejad, Nima, TU Wien, Institute of Computer ous measurement of individually varying product properties in early Technology, Gußhausstraße 27–29, 1040 Vienna, Austria; Jantsch, Axel, TU Wien, stages [4]. In view of these requirements, researchers have proposed Institute of Computer Technology, Gußhausstraße 27–29, 1040 Vienna, Austria (E-mail: axel.jantsch@tuwien.ac.at) various methods of intelligent sensing, self-organization, and self- 270 heft 3.2018 The Author(s) e&i elektrotechnik und informationstechnik L. C. Siafara et al. SAMBA – an architecture for adaptive cognitive control... ORIGINALARBEIT required for autonomy, robustness, and safety in the system. (iii) Fi- studied by Kaindl et al. [11]. An agent is defined as a combination nally, the third challenge regards the communication and decentral- of hardware and software components which represents itself and ized existence of information in a distributed system, operating in its relations to its environment in an explicit symbolic manner. The known or unknown environments. proposed system is used for self-configuration as well as monitoring In this paper, we propose an architecture designed to address and failure detection. However, no cause detection was considered in that work. these individual challenges in an integrated and comprehensive Self-monitoring has been implemented also as a hierarchical manner in order to meet the expectations of modern production agent in a Systems-on-Chip (SoC) [12]. The proposed structure fa- systems. Therefore, the proposed system aims at the dynamic ob- cilitates the process of monitoring parallel many-core SoCs. Cyber- servation of the environment, data abstraction, and distributed ne- Physical System-on-Chip is another platform where self-awareness gotiation. They serve each unit of the system in detecting the faults has been explored [13]. There, the authors described self-awareness and anomalies independently but collaboratively. Each unit uses self- as the ability of a system to monitor its own internal and external be- configuration to achieve dynamic clustering of the system for the havior in order to make appropriate decisions. The superiority of the purpose of communication between units. Finally, the system at- hereby proposed architecture of self-aware health monitoring over tempts to go beyond using the existing actions for known prob- the state of the art is its ability to interpret the reliability of the data, lems and to mitigate new anomalies and problems to guarantee the measure the confidence of its processes, use attention for more ef- safety of the whole system. ficient resource utilization, and perform predictions to provide more It should be born in mind that the proposed architecture is work information for the decision-making process. in progress and not yet completely implemented. We have sim- ulated the SAMBA architecture in the nxtStudio based environ- 2.2 Cognitive systems ment and verified its main functionality, which we present here. Flexible and reliable operation in many automated processes is We have implemented a portion of Self-Aware Health Monitoring achieved by the inclusion of the human operator in the loop [14]. (SAHM), namely health monitoring of an injective function black Cognitive abilities enable humans to solve problems under uncer- box using contextual information, which we tested on an AC- tainty and despite the changes in the environment and tasks. Al- motor case study [8]. Moreover, we have recently studied the util- though such capabilities are common in humans, they are rarely ity of SAMBA features in the context of hierarchical goal manage- found in industrial systems. The goal of cognitive architectures is ment and found encouraging preliminary results for their use in to define a framework for designing systems with human-like intel- autonomous anomaly mitigation, self-configuration and arbitration ligence; they provide a structure which enables a system to develop between conflicting goals [9]. Although we are encouraged by the over its lifetime by embedding the mechanisms of perception, rea- results so far, the main challenge will be to demonstrate that SAMBA soning, action, and learning [15]. Different cognitive architectures leads to a measurable improvement in overall reliability and opera- have been applied for realizing cognitive tasks, for example, SOAR tional efficiency of the production system. That is the target of an [16], LIDA [17] and ACT-R [18]. ongoing FFG funded project with TU Wien, Danube University Krems To address limitations related to incomplete sensing of the envi- (DUK), nxtControl, and AVL (Anstalt für Verbrennungskraftmaschi- ronment and inability to individually carry-out global tasks, cogni- nen List) as partners. tive systems often exhibit social behavior, which entails communica- This paper first reviews the state of the art in the areas of tion and cooperation or negotiation. Such an approach is adopted self-aware monitoring, cognitive systems and dynamic clustering in in cognitive radio networks, where multiple distributed sensors col- Sect. 2. In Sect. 3, it explains how these methods are integrated into laborate selectively to enhance their spectrum sensing performance the SAMBA architecture and presents the modules of the solution and the utilization of the radio frequency spectrum [19]. Distributed and their interfaces. Next, in Sect. 4, it goes through an exemplary cognitive systems have been studied also for the control of robots scenario with disturbances to elaborate how the system handles the [20], and building environmental control [21]. In industrial appli- problem and presents a high-level simulation of the system using cations, the cognitive production systems refer to highly intercon- IEC 61499 function blocks. Finally, it discusses potential benefits of nected devices with improved sensing, reasoning, learning and plan- the proposed architecture and draws conclusions. ning capabilities, which use knowledge-based and learning mod- els to assess and expand their capabilities [1]. The cognitive system 2. State of the art design paradigm provides a promising approach towards the dy- namic adaptation of the processes and the continuous optimization 2.1 Self-aware monitoring through learning by observation of the environment and reasoning Recent attempts to improve the efficiency, collaboration and re- for mitigating errors and finding improved operation strategies. silience of automated systems in industry highlight the importance A more recent cognitive architecture is the Simulation of Mental Apparatus (SiMA) [22], which has previously been applied for cog- of the self-awareness in such systems. Self-awareness enables a sys- nitive control in building automation [23]. SiMA focuses on func- tem to monitor itself and its own environment to assess its situation tions that generate human behavior and implements the underly- better and make more appropriate decisions. An architecture for ing mechanisms (e.g., drives, emotions) which drive a system to cyber-physical manufacturing system based on Industry 4.0 has been exhibit a certain behavior. This behavior is thereby defined by the proposed by Lee et al. [10]. The authors studied a unified 5-level ar- internal state of the system, which on its turn depends on the feed- chitecture. The first level deals with the data acquisition and then back returned by the environment on the system’s actions, rather the self-awareness is used in the second level to monitor the health than being explicitly defined by the environmental state. Although degradation of the system. In another study, self-representation for this approach requires a higher engineering effort in the beginning monitoring automation systems using a multi-agent structure was due to the need for implementing the underlying driving behavioral mechanism, it can achieve a lower system complexity since the states under which a behavior is exposed do not need to be defined ex- SAVE (Self-monitoring-based process Adaptation for quality assurance in het- erogeneous VErsatile manufacturing), funded by FFG under contract 864883. plicitly [24]. The SiMA theory is used as basis for the design of the Juni 2018 135. Jahrgang The Author(s) heft 3.2018 271 ORIGINALARBEIT L. C. Siafara et al. SAMBA – an architecture for adaptive cognitive control... cognitive module in SAMBA, which is discussed in detail in the next section. 2.3 Dynamic clustering A key to an efficient performance in distributed systems is commu- nication among system components. The concept of clustering was devised in conjunction with the need to adapt to high complex- ity and the dynamic nature of emerging manufacturing control sys- tems. It originated from the Holonic Manufacturing Systems (HMS) concept [25] and the MetaMorph approach of HMS [26], based on the idea of expanding problem-solving clusters. Each cluster is a dy- namically created community of intelligent system components that cooperate with each other to obtain enough information to solve a problem. Each member of a cluster has a set of algorithms which enable recognition of a problem and finding a solution. A flexible structure is one of the advantages of a distributed clustering con- cept that allows solving complex problems in a scalable manner. Clustering has found its use in a variety of industrial applications such as mitigating the complexity of production orders [27], knowl- Fig. 1. The architecture of SAMBA depicting the structure of one En- edge propagation of thermal modeling [28], ad-hoc wireless net- tity works [29] for optimizing the routing protocol [30], andinother concepts based on swarm intelligence [31, 32]. The proposed architecture uses the concept of distributed clus- 3.2 Self-aware knowledge extraction and representation tering for establishing relevant connections between independently SAHM aims at parallel data abstraction and health status analysis acting entities in the system. Similar to [27], it establishes connec- (anomaly detection) to provide the system with a suitable observa- tions between the system components according to the production tion required for self-awareness [33]. The data is abstracted in a order structure. In this case, however, the goal (rather than decision- normal situation when the sensors’ functionality is correct. How- making) is more relevant to knowledge propagation [28] and there- ever, when some of the sensors are faulty, the intention is to request fore, the architecture presents a novelty over the current state of the equivalent data from adjacent entities. The health status analysis the art. approach mostly relies on the self-observation, however, sometimes negotiation with other entities increases the knowledge of an entity 3. Proposed architecture regarding an anomaly. Therefore, the proposed self-aware system Figure 1 illustrates the architecture of SAMBA and its major mod- monitors various health parameters in a distributed manner. These ules. The logical unit of an entity is an Autonomous Cooperating parameters are used to classify the health status of the entity into a Object (ACO). The ACO learns locally the specifics of the environ- predefined state. The input parameters are the data of the sensors ment and the actions it can take. Further, it exhibits social behavior, that have been abstracted. These sensors may belong to an entity since it interacts with other ACOs located in the same environment. itself or to other entities. To decide on its agenda, it takes into consideration its own goals, The goal of abstraction is to reduce the size of data as well as the environmental context and the requests from other ACOs. The to extract knowledge which is beneficial for the further processes global behavior emerges out of the interactions among the ACOs. in SAHM, the internal manager, or other entities. It deals with the sensor data and also other specifications about the hardware part 3.1 System architecture of the entity, including the nominal range and the reliability of each Each ACO consists of three SAMBA-specific functional units in addi- component. Abstraction has several functional states, e.g., abstrac- tion to the operation module: Operation Module serves as the inter- tion of sensor data, providing a response to the internal manager’s face to the hardware. It connects to the plant controller and forms a request for specific data as well as providing specific data to the major part of the by providing symbolized sensor data and receiving external manager, or processing the data provided from another ex- actuator ACO commands. Often, this unit sits on the legacy con- ternal manager. troller. SAHM monitors the system operation and in case of anoma- The goal of health status analysis is to detect the anomaly (as lies, it proposes possible causes. To this end, SAHM performs fault well as its characteristics, causes, and effects), and to inform the diagnosis and data abstraction. As each ACO is observing only its internal manager. It also deals with the requests from the external local environment, other ACOs can request or may be requested manager regarding the health and operational status of the entity. Health status analysis includes anomaly detection, anomaly specifi- to provide further information through the external manager. Inter- cation, cause diagnosis and prediction blocks. The abstracted data nal Manager is the decision-making entity of the system. It receives first enters the anomaly detection. The type of anomaly and the health status information from SAHM and external requests from value of the reliability are the outputs which are forwarded to the other ACOs through the external manager. At a certain state, there other three blocks. The detected anomaly triggers the anomaly spec- might be conflicting action requests and the internal manager shall ification block which estimates the features of this anomaly, such decide which goal is prioritized. External Manager provides an im- as the rate of wear-out deterioration. Cause diagnosis detects the plicit connectivity among the relevant ACOs via distributed cluster- reason(s) of the anomaly. This block requires the abstracted data, ing based on the production order structure and the physical layout type of anomaly and features of the anomaly (if they exist) as well of the shop floor. It facilitates transparent negotiations between one as complementary information from other ACOs. If the health status ACO and other ACOs in the cluster. In the rest of this section, we analysis suspects the existence of an external cause, a data request is discuss the main functionalities of each module. 272 heft 3.2018 The Author(s) e&i elektrotechnik und informationstechnik L. C. Siafara et al. SAMBA – an architecture for adaptive cognitive control... ORIGINALARBEIT sent to the external manager, which can ask for the data and health have not been activated for a long time. Decisions of the Internal status of the other involved ACOs. Another output of this block is Manager (IM) are evaluated and their effects are updated over time the new value of specification for the reliability of the components in order to account for the changing dynamics of the environment, which is transferred to the abstraction block. Finally, the prediction therefore, maintaining an implicit model of the environment. block receives the abstracted data, type of anomaly, its features, and cause(s), and predicts the future effect of the anomaly on the entity 3.4 Adaptive collaboration or the system. To compensate for the lack of a global overview due to the dis- The frequently updated reliability information of the data and tributed nature of the proposed system, the concept of dynamic components provided from MES as well as the confidence value clustering is applied. Dynamic clustering introduces a flexible way at the end of each step are all interpreted as metrics to measure to integrate global objectives while allowing encapsulation of core the level of uncertainty. Different approaches are used depending functionality of the ACO, hence, providing transparency for the on where the uncertainty is flagged. For example, if the abstrac- decision-making. The challenge is to form clusters regarding the tion unit faces unreliable data, a negotiation with the neighboring functional requirements of the shop floor and manufacturing in- ACO(s) starts to request data from them. A self-aware Multiple Clas- structions. Currently, there is no methodology to formalize produc- sifier System (MCS) based on the rankings of each individual clas- tion order ontology in relation to the anomalies. Instead of pre- sifier similar to [34] can be used to improve the performance of defined rules of the existing solutions, the External Manager (EM) anomaly detection and cause diagnosis through analysis of a col- defines dynamic methods using learning algorithms (e.g., neural lecting of results and thereby choosing the optimal one to reduce networks, and reinforcement learning) to automatically discover the the uncertainty. Cause detection can be built using a MCS whose cluster formation rules based on the production order, the physical inputs are the type of anomaly, meta-data and anomaly parameters. layout of the shop floor, and network communication structure. A Multiple learning algorithms interpret the inputs and provide the re- set of generic classifiers which operate on non-application-specific sult with a value of confidence. If an external cause is probable to characteristics are defined and used by the EM to determine es- obtain relevant external data, negotiation with other ACOs is initial- sential connections in the cluster and establish weighted links be- ized. The new external information is fed as an input to the system tween ACOs based on the prioritized set of control parameters. and a new analysis starts. These weighted connections are updated at runtime using back- propagation according to feedback from the evaluation of the com- 3.3 Cognitive decision-making munication and the execution of decisions. To select the most suitable actions with regard to the goals of an As a result, the system achieves implicit connectivity among the ACO, decision-making takes place on the basis of data inputs from relevant ACOs and facilitates transparent negotiations between the SAHM and the external manager. SAHM provides information SAHM and the internal manager of different ACOs in a cluster. The regarding the current state of the ACO and its environment, whereas External Manager bases its technique on a non-deterministic rele- the external manager provides information regarding the state of vance of connection between individual ACOs regarding a particular other ACOs and their environment. Two input types are defined: event. The outputs of the External Manager are used to check possi- drives and percepts. Drives are the (initial) motivations of the system ble implications of local decisions on other ACOs in the community. and therefore, the source of the system goals. Their intensity rep- By adjusting the relevance of the connections over time, such de- resents the deviation from a desired goal state: the higher a drive pendencies are dynamically adapted based on the consequences of intensity, the more urgent the goal represented by this drive. Sym- actions. bols from the environment constitute the percepts of the ACO. Once percepts are generated for the current state, similar states from the 3.5 System dynamics and interface memory are activated. Emotions are the internal evaluation mech- Communication between the units of an ACO and also between dif- anism of the system and are generated based on the current state ferent ACOs of a cluster is necessary to achieve the functionality of of drives and the stored activated memories from the past. All these the larger system described in Fig. 2. This figure highlights the wiring processes form part of the primary process of the system, which is and content of data exchanges between units in Fig. 1. The environ- characterized by the lack of reasoning functions. The primary pro- ment consists of the yellow blocks on the top, exchanging informa- cess regulates the reactive behavior of the system and proposes tion with the Operation Module in the ACO, and other ACOs shown actions that are able to solve urgent problems in a reactive man- at the bottom. Types of information transferred between units inside ner [35]. the ACO are also specified and depicted in Fig. 2. Finally, the ACO In the secondary process, social rules which reward or penal- is connected to the External Manager of other ACOs, which form ize a behavior are tested on the current state; they represent user the other part of the environment. Note that this figure illustrates preferences and guide the policies that the system should comply the minimum possible connections in the proposed system. Each re- with them. The rewards, which represent the external evaluations, quest and its reply are labeled by an ID to simplify the process of along with the emotions, which are the internal evaluations, en- handling them. To focus on the main system functionality, it is as- rich the current perceived state. The enriched perceived state and sumed that no communication errors occur. the drives, which indicate goal priorities, are the inputs to the goal The connection between the Internal Manager and the Operation selection process, where it is decided which goal the system shall Module is established for transferring the estimated set-points after pursue. Next, sequences of states (episodes) similar to the current the decision-making process. There is only a one-directional data episode are activated using case-based reasoning. From the acti- flow from the Operation Module to SAHM to push all data from vated episodes, sequences of actions (policies) that managed to re- sensors to SAHM. duce the drive intensities in the past are then picked as potential Several communication messages are separately sent from SAHM options. The policy with the best performance is used to select the current action to execute. The new episode is saved in the memory to the Internal Manager including a value representing the health by the learning process, which is responsible for the tasks of updat- status of the ACO, information on anomalies and their cause, ex- ing the memory with new episodes while it removes episodes that tra data in response to Internal Manager’s requests, and the action Juni 2018 135. Jahrgang The Author(s) heft 3.2018 273 ORIGINALARBEIT L. C. Siafara et al. SAMBA – an architecture for adaptive cognitive control... conveyor sections supplying two robotic arms with objects. includes three different types of entities, i.e., the conveyor sections, which transport the products, the product sorters, which sort the products by type for further processing, and the robotic arms, which process the products. Each entity consists of the hardware part (i.e., mo- tor, sensors, and other equipment), and the logic unit (the ACO), which monitors and, if necessary, adjusts its operation. The system monitors the existence of all items in its environment. If an ACO is expecting to receive an item at a specific time but it does not, the system shall try to find the root cause of the event. To this end, it uses the observations from entities responsible for previous (or next) tasks in the process and analyzes the available information to find out if the item is missing or the sensor is not working. Afterwards, the system needs to decide whether it can adjust the settings and resolve the problem, or the operator needs to be notified. The following tasks take place: 1. The SAHM detects the event and tries to diagnose the problem. That is, to find out whether the problem is caused by a faulty sen- sor or a missing product. It informs the internal manager about the event. 2. The external manager, upon notification from the internal man- ager, asks for sensor information from the previous/next ACOs in the process. 3. The SAHM receives the information from the other related ACOs, and analyzes all information to find out if the item is missing or the sensor is not working. This action will continue in a sequence until the source of the problem (product missing/stuck or sensor Fig. 2. Communication flow within the ACO and with other ACOs of failure) is found. the cluster 4. If the problem is due to erroneous sensor data, the SAHM notifies the internal manager for this event to take the necessary actions to handle the problem. proposal. The action proposal responds to a request made by the 5. If the problem is that the product is missing, due to unsynchro- Internal Manager and indicates the operational range within which nized speed settings of the source entities, the SAHM notifies the ACO can adapt its operational settings. Data and action requests the internal manager and the latter takes the necessary actions are sent from the Internal Manager to SAHM. Action requests are to handle the problem. sent to SAHM upon receipt of an action request from the external 6. If the problem is that the product is lost, then the internal man- manager, in order to ask for the nominal operational range of the ager informs the human operator about the fault and asks for ACO. intervention. The Internal Manager establishes a connection to the External Manager when the decision it makes impacts the operational pa- rameters of another ACO. An action request is also sent when the 4.2 Feasibility verification of communication architecture Internal Manager decides to change an operation parameter and the To verify the feasibility of the communication architecture, we ran negotiation needs to be initiated. Moreover, in response to an estab- a simulation experiment which we present in the rest of this sec- lished negotiation from other ACOs, the Internal Manager sends a tion. The simulation was performed in the nxtStudio runtime envi- confirmation message to the requesting ACO in case of an agree- ronment which uses the IEC 61499 standard. The standard is based ment. on the event-driven concept of function blocks. Function blocks are The messages from SAHM to the External Manager include data independently acting components that encapsulate local function- requests when SAHM detects an anomaly and needs to contact ality and communicate with each other via events and associated another ACO for additional information. Similar messages are ex- data. Each block implements a set of algorithms and triggers events changed in the opposite direction too. The communication mecha- upon completion. The main benefit of the IEC 61499 model lies in nisms between External Managers of different ACOs are similar to its flexibility to integrate complex systems and adapt them even at the ones between the External Manager and the Internal Manager. runtime. In addition, External Managers exchange data requests to establish The main goal of the simulation was to establish a communication clusters. infrastructure for testing the synergy of the individual SAMBA com- ponents. For each node in the production line of the use case (Fig. 3), 4. Exemplary use case a corresponding ACO was created. The general structure of an ACO is showninFig. 4; the ACO is represented as a composite function 4.1 Definition/description block with SAHM, IM, EM and OM as basic function blocks. Event To explain how the system behavior emerges out of the interactions inputs and outputs with associated data were defined and the com- among the individual components of the ACOs as well as through munication algorithms were implemented. The resulting simulation communication with other ACOs, consider an exemplary scenario with detection of a missing product. Figure 3 illustrates a poten- provides a flexible environment for testing the underlying algorithms tial configuration of the system; an assembly line consisting of four of SAMBA that can be used as an application independent runtime 274 heft 3.2018 The Author(s) e&i elektrotechnik und informationstechnik L. C. Siafara et al. SAMBA – an architecture for adaptive cognitive control... ORIGINALARBEIT Fig. 3. Shop floor configuration in the missing product use case Fig. 4. ACO function-block structure environment. Due to the modular nature of the IEC 6149 standard, The design of SAMBA addresses following challenges which mod- a variety of use cases and algorithms can be integrated and tested ern industrial production process faces: to analyze the optimal configuration of algorithms and parameters 1. Increased adaptivity and reduced engineering effort: to interpret for a particular application. The simulation results showed the sat- the context of operation with minimum a priori knowledge avail- isfying level of scalability and flexibility of the proposed architecture able, semantic enhancement of the available information is used. and its ability to integrate required algorithms for data acquisition This is enabled by self- and context-awareness concepts such as and decision-making. confidence, attention, data-reliability, and history, combined with intelligent data analysis, such as data abstraction and scattered- 5. Discussion data fusion. SAMBA is an architecture for CPPS that monitors the production 2. Quality assurance and resilience enhancement through fault process and reacts to deviations, either through automatic compen- diagnosis and prognosis: in a distributed system no single sations or by informing the operator. It is designed to operate as a (sub)system knows the complete state of the overall system. middleware on the top of existing (legacy) systems adding a layer Communication with other ACOs takes place to enhance local of intelligence to the CPPS. Intelligence in this specific context is information, and to increase confidence about local knowledge. defined as the ability of the system to be context-aware and self- This also helps the system to obtain awareness of the bigger aware, to be proactive and social, acting within a certain degree of context in order to detect, analyze, predict, and mitigate errors, autonomy in a changing environment. To this end, SAMBA builds faults, and failures. upon the concept of the ACO, as presented above. The behavior of 3. Enhanced autonomy and intelligence for mitigation of anomalies the large system emerges out of the cooperation of individual ACOs, in operation: in complex manufacturing systems (such as CPPSs) which dynamically form clusters and collaborate on demand while thorough and precise modeling of the system and its environ- pursuing their own goals as well as the goals of the global system. To ment is challenging. Cognitive systems are adept at making deci- compensate for the lack of central supervision, on one hand, ACOs communicate with each other to enhance their knowledge and un- sions efficiently despite the lack of a complete or precise model. derstand the context of their observations. On the other hand, they Cognitive decision-making improves the efficiency of the system negotiate to align their actions in order to efficiently achieve the in using extracted knowledge about each ACO, its neighbors in global system goals. the cluster, the overall system, and possible courses of action, to Juni 2018 135. Jahrgang The Author(s) heft 3.2018 275 ORIGINALARBEIT L. C. Siafara et al. SAMBA – an architecture for adaptive cognitive control... re- or pro-actively change the CPPS, regarding the timing con- 5. Zhang, Y., Qian, C., Lv, J., Liu, Y. (2017): Agent and cyber-physical system based self- organizing and self-adaptive intelligent shopfloor. IEEE Trans. Ind. Inform., 13(2), 737– straints at hand. Especially by mitigation of failures, degraded health, or anomalies, it ensures the quality of the product and 6. Park, H.-S., Tran, N.-H. (2012): An autonomous manufacturing system based on swarm process in adaptive heterogeneous manufacturing systems. of cognitive agents. J. Manuf. Syst., 31(3), 337–348. 4. Dynamic clustering based on environment discovery: dynamic 7. Monostori, L. (2014): Cyber-physical production systems: roots, expectations and r&d cluster building concepts which take into account the produc- challenges. Proc. CIRP, 17, 9–13. 8. Götzinger, M., TaheriNejad, N., Kholerdi, H. A., Jantsch, A. (2017): On the design of tion orders parsing allow effective negotiation and propagation context-aware health monitoring without a priori knowledge; an ac-motor case-study. of mitigation measures. They also enable representation of the In 2017 IEEE 30th Canadian conf. on electrical and comp. eng. (CCECE) (pp. 1–5). overall production environment beyond its locally accessible in- 9. Jantsch, A., Anzanpour, A., Kolerdi, H., Azimi, I., Siafara, L. C., Rahmani, A. M., Taher- formation. inejad, N., Liljeberg, P., Dutt, N. (2018): Hierarchical dynamic goal management for IoT systems. In Proc. of the IEEE int. symp. on quality electronic design, USA. SAMBA aims to introduce a new design paradigm in the industrial 10. Lee, J., Bagheri, B., Kao, H.-A. (2015): A cyber-physical systems architecture for indus- try 4.0-based manufacturing systems. Manufac. Lett., 3, 18–23. environment which, thanks to the generic nature of the proposed 11. Kaindl, H., Vallée, M., Arnautovic, E. (2013): Self-representation for self-configuration architecture, allows integration of legacy systems in manufacturing and monitoring in agent-based flexible automation systems. IEEE Trans. Syst. Man with considerably less effort compared to the state of the art. The Cybern., 43(1), 164–175. generic nature of the architecture makes it possible to apply it inde- 12. Guang, L., Plosila, J., Isoaho, J., Tenhunen, H. (2010): Hierarchical agent monitored pendent of the specifics of the entities it controls. It is expected to parallel on-chip system: a novel design paradigm and its formal specification. Int. J. Embed. Real-Time Commun. Syst., 1(2), 85–105. reduce the downtime and frequency of maintenance interventions 13. Sarma, S., Dutt, N., Gupta, P., Nicolau, A., Venkatasubramanian, N. (2014): On-chip by health monitoring, predictive analysis and mitigation of behav- self-awareness using cyberphysical-systems-on-chip (CPSOC). In Proc. intl. conf. on ioral deviations. Next steps in the development of SAMBA include hardware/software codesign and system synthesis (p. 22). New York: ACM. the detailed design of the algorithms for the modules of the ar- 14. Zhao, Y. F., Xu, X. (2010): Enabling cognitive manufacturing through automated on- chitecture and testing of the emergent system performance. In the machine measurement planning and feedback. Adv. Eng. Inform., 24(3), 269–284. 15. Vernon, D. (2017): Two ways (not) to design a cognitive architecture. Cognitive Robot planned follow-up project, SAVE, the challenge will be to demon- Architectures, 42. strate – quantitatively – that SAMBA improves the overall system 16. Laird, J. E. (2012): The Soar cognitive architecture. Cambridge: MIT Press. performance in terms of reliability and operation efficiency. 17. Snaider, J., McCall, R., Franklin, S. (2011): The lida framework as a general tool for agi. In Int. conf. on AGI. Berlin: Springer. 18. Taatgen, N. A., Lebiere, C., Anderson, J. R. (2006): Modeling paradigms in act-r. Cog- 6. Conclusion nition and multi-agent interaction: from cognitive modeling to social simulation, pp. In this paper, we presented SAMBA, an architecture for versatile 29–52. heterogeneous CPPS with focus on predictive analysis, autonomous 19. Yucek, T., Arslan, H. (2009): A survey of spectrum sensing algorithms for cognitive health monitoring, and error mitigation in industrial manufacturing radio applications. IEEE Commun. Surv. Tutor., 11(1), 116–130. processes. The synergies provided by these functionalities reduce 20. Kawamura, K., Peters, R. A. II., Bodenheimer, R. E., Sarkar, N., Park, J., Clifton, C. A., Spratley, A. W., Hambuchen, K. A. (2004): A parallel distributed cognitive control system reaction time and engineering efforts through adjustment system for a humanoid robot. Int. J. Humanoid Robot., 1(01), 65–93. of production steps to compensate for the deviations. SAMBA in- 21. Kollmann, S., Siafara, L. C., Schaat, S., Wendt, A. (2016): Towards a cognitive multi- creases the ability of the system to assure the quality of the product agent system for building control. Proc. Comput. Sci., 88, 191–197. and the process by intelligently adapting to rapidly changing envi- 22. Schaat, S., Wendt, A., Kollmann, S., Gelbard, F., Jakubec, M. (2015): Interdisciplinary development and evaluation of cognitive architectures exemplified with the SIMA ap- ronments. The architecture is mainly developed for distributed CPPS, proach. In EAPCogSci. however, its generic and modular design enables its future applica- 23. Zucker, G., Habib, U., Blöchle, M., Wendt, A., Schaat, S., Siafara, L. C. (2015): Building tion to different industrial systems with minimum effort for adapta- energy management and data analytics. In EDST, 2015 int. symposium (pp. 462–467). tion. 24. Wendt, A., Kollmann, S., Siafara, L., Biletskiy, Y. (2018): Usage of cognitive architec- tures in the development of industrial applications. In ICAART. 25. Van Brussel, H., Wyns, J., Valckenaers, P., Bongaerts, L., Peeters, P. (1998): Reference Acknowledgements architecture for holonic manufacturing systems: Prosa. Comput. Ind., 37(3), 255–274. Open access funding provided by TU Wien (TUW). We gratefully 26. Kühnle, H. (2009): Distributed manufacturing: paradigm, concepts, solutions and ex- acknowledge the financial support provided to us by the BMVIT/FFG amples. Berlin: Springer. under the program ICT of the Future in the project SAMBA (contract 27. Babiceanu, R. F., Chen, F. F. (2006): Development and applications of holonic manu- facturing systems: a survey. J. Intell. Manuf., 17(1), 111–131. number: 855426). 28. Bratukhin, A., Nagy, A., Mahmood, A. (2015): Distribution of control functionality in energy-aware industrial building environment. In WFCS, IEEE world conf. (pp. 1–8). Open Access This article is distributed under the terms of the Creative New York: IEEE Press. Commons Attribution 4.0 International License (http://creativecommons.org/ 29. Abbasi, A., Younis, M. (2007): A survey on clustering algorithms for wireless sensor licenses/by/4.0/), which permits unrestricted use, distribution, and reproduc- networks. Comput. Commun., 30(14), 2826–2841. tion in any medium, provided you give appropriate credit to the original au- 30. Liu, X. (2012): A survey on clustering routing protocols in wireless sensor networks. thor(s) and the source, provide a link to the Creative Commons license, and Sensors, 12(8), 11113–11153. indicate if changes were made. 31. Saleem, M., Di Caro, G. A., Farooq, M. (2011): Swarm intelligence based routing protocol for wireless sensor networks: survey and future directions. Inf. Sci., 181(20), References 4597–4624. 32. Kennedy, J. (2006): Handbook of nature-inspired and innovative computing (pp. 187– 1. Zaeh, M. F., Reinhart, G., Ostgathe, M., Geiger, F., Lau, C. (2010): A holistic approach 219). Berlin: Springer. for the cognitive control of production systems. Adv. Eng. Inform., 24(3), 300–307. 33. Taherinejad, N., Jantsch, A., Pollreisz, D. (2016): Comprehensive observation and its 2. Shea, K. (2010): The cognitive factory. role in self-awareness – an emotion recognition system example. In Conf. on comp. 3. European Factories of the Future Research Ass. et al. (2013): Factories of the future: science and inf. sys. Multi-annual roadmap for the contractual ppp under horizon 2020. Pub. office of the 34. Kholerdi, H. A., TaheriNejad, N., Jantsch, A. (2018): Enhancement of classification of Eur. Union: Brussels. small data sets using self-awareness – an iris flower case-study. In Proc. of the int. 4. Armengaud, E., Sams, C., von Falck, G., List, G., Kreiner, C., Riel, A. (2017): Industry symp. on circuit and systems (ISCAS), Florence, Italy. 4.0 as digitalization over the entire product lifecycle: opportunities in the automo- 35. Zucker, G., Wendt, A., Siafara, L., Schaat, S. (2016): A cognitive architecture for build- tive domain. In Eur. conf. on software process improvement (pp. 334–351). Berlin: ing automation. In Ind. electronics society, IECON 2016 (pp. 6919–6924). New York: Springer. IEEE Press. 276 heft 3.2018 The Author(s) e&i elektrotechnik und informationstechnik L. C. Siafara et al. SAMBA – an architecture for adaptive cognitive control... ORIGINALARBEIT Authors current research focus is in the area of distributed control sys- Lydia Chaido Siafara tems, artificial intelligence, machine learning and modeling in in- received her degree in electrical engineer- dustrial and building automation with the focus on efficient energy ing and computer technology from Aristo- use. tle University of Thessaloniki, Greece, and the master’s degree in building science and technologies from Vienna University of Tech- Nima Taherinejad nology (TU Wien) in 2014. Since 2015 she is a Ph.D. graduate of The University of British has been working as a project assistant at Columbia, Vancouver, BC, Canada. He is cur- TU Wien, where she was responsible for the rently a Post-doctorate University Assistant at KORE project, dealing with the application of TU Wien, Vienna, Austria, where his main ar- human-inspired cognitive systems in building automation. Her work eas of research include systems on chip, self- is concerned with the development of intelligent agents and their awareness in cyber-physical systems, embed- application in the field of automation systems. Her research interests ded systems, and robotics. He was the gen- include artificial intelligence, cognitive systems and machine learn- eral chair and TPC chair of MobiHealth 2017, ing. chair of the AUTO21 HQP Advisory Commit- tee and a member of board of directors of AUTO21 Center of Excel- Hedyeh Kholerdi lence. He has authored a book and published in/served as a reviewer received her master’s degree in the field of for various journals and conferences. Dr. Taherinejad has received Image Processing from Babol University of several awards and scholarships from universities and conferences Technology, Iran. Her master’s thesis is titled he has attended. as “Drowsiness detection using image pro- cessing technique inspired by the human vi- Axel Jantsch sual system”. She joined TU Wien, Institute received the Dipl.-Ing. and Dr. techn. degrees of Computer Technology (ICT) as a university from TU Wien, Vienna, Austria, in 1988 and assistant for Embedded Systems and Online 1992, respectively. From 1997 to 2002, he VHDL Programming courses. She was also a was an Associate Professor with the KTH project assistant working on self-aware health monitoring in scalable Royal Institute of Technology, Stockholm, distributed systems. The main focus was on research in requirement Sweden, where he was also a Full Professor of creating a contextual awareness in a distributed autonomous sys- of Electronic Systems Design from 2002 to tem using learning approaches. 2014. Since 2014, he has been a Professor with the Institute of Computer Technology, Aleksey Bratukhin TU Wien. He has authored over 300 articles and one book in the received his master’s degree at Perm Techni- areas of VLSI design and synthesis, HW/SW codesign and cosyn- cal University in Russia in 1998. From 2000 thesis, networks-on-chip, and self-awareness in Cyber-Physical Sys- to 2006 he worked at Vienna University of tems. Technology with the focus on software agent systems and distributed control in respect to vertical integration in the area of plant au- tomation, where he received his doctorate degree in 2006. In that year he joined the Center for Integrated Sensor Systems. His Juni 2018 135. Jahrgang The Author(s) heft 3.2018 277

Journal

e & i Elektrotechnik und InformationstechnikSpringer Journals

Published: Jun 4, 2018

References

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off