Autonomy in production logistics: Identi“cation ...Autonomy in production logistics:...
Transcript of Autonomy in production logistics: Identi“cation ...Autonomy in production logistics:...
Robotics and Computer-Integrated Manufacturing ] (]]]]) ]]]–]]]
Autonomy in production logistics: Identification,characterisation and application
Katja Windt, Felix Bose, Thorsten Philipp�
Bremen Institute of Industrial Technology and Applied Work Science at the University of Bremen, Bremen, Germany
Abstract
Competitive enterprises have to react fast and flexible to an increasing dynamic environment. To achieve the ability to adapt on these
new requirements autonomous cooperating logistic processes seem to be an appropriate method. In order to prove in which case
autonomously controlled processes are more advantageous than conventionally managed processes, it is essential to specify what is
exactly meant with autonomous control, how autonomous control does differ from conventional control and how the achievement of
logistic objectives in autonomously controlled systems can be estimated and compared to the achievement of objectives in conventionally
controlled systems. This paper introduces a general definition of autonomous control as well as a definition in the context of engineering
science and its meaning in a logistics context. Based on this, a catalogue of criteria is developed to ensure the identification of
autonomous cooperating processes in logistic systems and its distinction to conventionally controlled processes. To demonstrate this
catalogue, its criteria and the concerning properties are explained by means of an exemplary shop-floor scenario.
r 2007 Published by Elsevier Ltd.
Keywords: Autonomous control; Logistics; Production systems; Production planning and control
1. Initial situation and call for action
Over the past years, an increase in structural anddynamic complexity of production and logistics systemscould be observed. This development is caused by diversechanges, for example, short product life cycles as well as adecreasing number of lots with a simultaneously risingnumber of product variants and higher product complexity[1]. As a result, new demands were placed on competitivecompanies, which cannot be fulfilled with conventionalcontrolling methods. Conventional production systems arecharacterised by central planning and controlling pro-cesses, which do not allow fast and flexible adaptation tochanging environmental influences. Establishing autono-mous cooperating logistics processes seems to be anappropriate method to meet these requirements. The ideaof autonomous cooperating logistic processes is to developdecentralised and heterarchical planning and controllingmethods in contrast to existing central and hierarchical
aligned planning and controlling approaches [2]. Autono-mous decision functions are shifted to logistic objects. Inthe context of autonomous control logistic objects aredefined as material items (e.g. part, machine, conveyor) orimmaterial items (e.g. production order) of a networkedlogistic system, which have the ability to interact with otherlogistic objects of the considered system. Autonomouslogistic objects are able to act independently according totheir own objectives and to navigate through the produc-tion network themselves. The autonomy of logistic objectsis possible since recent developments by information andcommunication technologies (ICT), for example radiofrequency identification (RFID) for identifying, globalpositioning system (GPS) for locating or universal mobiletelecommunications system (UMTS) for communicating oflogistic objects [3].These new approaches are currently investigated within
the Collaborative Research Center ‘‘Autonomous LogisticProcesses—A Paradigm Shift and its Limitations’’ at theUniversity of Bremen, which deals with the implementa-tion of autonomous control as a new paradigm forlogistic processes [4]. The intention of this paper is to
ARTICLE IN PRESS
www.elsevier.com/locate/rcim
0736-5845/$ - see front matter r 2007 Published by Elsevier Ltd.
doi:10.1016/j.rcim.2007.07.008
�Corresponding author.
E-mail address: [email protected] (T. Philipp).
Please cite this article as: Windt K, et al. Autonomy in production logistics: Identification, characterisation and application. Robot Comput Integr
Manuf (2007), doi:10.1016/j.rcim.2007.07.008
Published in: International Journal of Robotics and CIM, Pergamon Press Ltd, 2007.
explain what is meant by autonomous control and to showpotentials in logistic systems, particularly in productionsystems.
Therefore, a definition of the term autonomous controlis given. Based on the main statement of this definition, acatalogue of criteria is developed in order to identifyautonomous cooperating logistic processes and to empha-sise how conventionally managed and autonomous logis-tics processes differ. The criteria and its individualproperties are explained on a concrete object of investiga-tion by a scenario of a manufacturing system. Inconclusion further research activities concerning evaluationof autonomous control are presented.
2. Definition of autonomous control
The vision of autonomous cooperating logistics pro-cesses emphasizes the transfer of qualified capabilities onlogistic objects as explained above. According to thesystem theory, there is a shift of capabilities from the totalsystem to its system elements [5]. By using new technologiesand methods, logistic objects are enabled to renderdecisions by themselves in a complex and dynamicallychanging environment. Based on the first results of thework in the context of the Collaborative Research Centre(CRC) 637 ‘‘Autonomous Cooperating Logistic Pro-cesses—A Paradigm Shift and its Limitations’’ at theUniversity of Bremen [6], autonomous control can bedefined as follows:
Autonomous Control describes processes of decentra-lized decision-making in heterarchical structures. Itpresumes interacting elements in non-deterministicsystems, which possess the capability and possibility torender decisions independently.
The objective of Autonomous Control is the achieve-ment of increased robustness and positive emergence ofthe total system due to distributed and flexible copingwith dynamics and complexity. [7]
Based on this global definition of the term autonomouscontrol a definition in the context of engineering sciencewas developed, which is focussed on the main tasksof logistic objects in autonomously controlled logisticssystems:
Autonomous control in logistics systems is characterisedby the ability of logistic objects to process information,to render and to execute decisions on their own.
For a better understanding, the main statements of thegiven definitions of autonomous control such as decen-tralized decision-making in heterarchical systems, systemelements ability of interaction as well as non-deterministicsystem and positive emergence, are described and discussedbelow.
2.1. Decentralised decision-making in heterarchical systems
One feature of autonomous control is the capability ofsystem elements to render decisions independently. Auton-omy regarding decision-making is enabled by the alignmentof the system elements in the form of a heterarchicalorganisational structure [8]. Therefore, decentralisation ofthe decision-making process from the total system to theindividual system elements is a typical criterion ofautonomous control. Each system element represents adecision unit, which is equipped with decision-makingcompetence according to the current task [9]. Due to thefact that decision-making processes are purposeful, accord-ing to the decision theory, each system element in anautonomously controlled system is characterised by target-oriented behaviour. Global objectives, for example, pro-vided by the corporate management, can be modifiedindependently by the system elements in compliance withits own prioritisation. For example, the objective of highdue-date punctuality can be replaced in favour of highmachine utilization by the machine itself. Thus, theobjective system of single elements is dynamic becauseof its ability to modify objects prioritisation over time,i.e. during the production process.
2.2. System element’s ability of interaction
Decentralised decision-making processes require theavailability of relevant information for the system ele-ments. Consequently, the capability of system elements tointeract with others is a mandatory condition and thus oneconstitutive characteristic of autonomous control. Theability of interaction can accomplish different valuesdepending on the level of autonomous control. Theallocation of data, that other autonomous logistic objectscan access, represents a low level of autonomous control.Communication, i.e. bi-directional data exchange betweenautonomous logistic objects, and coordination, that meansthe ability of autonomous logistic objects to cooperate andcoordinate activities of other objects, stands for a higherlevel of autonomous control.
2.3. Non-deterministic system behaviour and positive
emergence
In accordance with the above-mentioned definition, themain objective of autonomous control is the achievementof increased robustness and positive emergence of the totalsystem due to a distributed and flexible coping withdynamics and complexity. Non-determinism means thatdespite precise measurement of the system status andknowledge about all influencing variables of the system, noforecast of the system status can be made. Knowledge of allsingle steps between primary status and following status isnot adequate enough to describe the transformationcompletely [10]. Thus, a fundamental criterion of autono-mous control is that for same input of initial values, there
ARTICLE IN PRESSK. Windt et al. / Robotics and Computer-Integrated Manufacturing ] (]]]]) ]]]–]]]2
Please cite this article as: Windt K, et al. Autonomy in production logistics: Identification, characterisation and application. Robot Comput Integr
Manuf (2007), doi:10.1016/j.rcim.2007.07.008
are different possibilities for transition in a followingstatus. As already explained, decentralisation of decision-making processes to the system elements leads to a higherflexibility of the total system, because of the ability to reactimmediately to unforeseeable, dynamic influencing vari-ables. In this way, autonomous control can lead to a higherrobustness of the overall logistic system. Furthermore,positive emergence is the main objective of autonomouscontrol. Emergence stands for development of newstructures or characteristics by concurrence of simpleelements in a complex system. Positive emergence meansthat the concurrence of single elements leads to a betterachievement of objectives of the total system than it isexplicable by considering the behaviour of every singlesystem element. That means related to the context ofautonomous cooperating logistic processes, that autono-mous control of individual logistic objects (e.g. machines,parts, orders) enables a better achievement of objectives ofthe total system than can be explained by consideration ofthe decentralised achievement of objectives (e.g. higher rateof on-time delivery, lower delivery times) of every singlelogistic object.
3. Derivation of a catalogue of criteria
The identification of autonomous cooperating logisticobjects requires a dissociation from conventionally mana-ged logistic objects. The definition of autonomous controlexplained in the preceding chapter describes the maximumlevel of imaginable autonomous control. Thus, all systemelements in an absolutely autonomous controlled systemare able to interact with other system elements and to
render decisions on the basis of an own, decentralisedtarget system. In general, logistics systems probablecontain both conventionally managed and autonomouslycontrolled elements and sub-systems, respectively. Further-more, it is assumed that there are different levels ofautonomous control, which is called level of autonomouscontrol. For example, one part of a production lot could beable to coordinate each production stage of the lot whichrepresents a high level of autonomy, meanwhile other partsonly allocate data regarding their processing states.Consequently, the latter mentioned case shows a lowerlevel of autonomy.Based on the definition of autonomous control in the
context of engineering science its main characteristics arethe ability of logistic objects to process information and torender and execute decisions. Each characteristic can beassigned to different layers of work in an enterprise. Inaccordance with Ropohl [11], different layers of work canbe classified in organisation and management, informaticsmethods and I&C technologies as well as in flow ofmaterial and logistics, each concerning decision, informa-tion and execution system. Fig. 1 presents the assignmentof the characteristics to the system layers, illustrates theircorrelations and introduces the main criteria of autono-mous control.The decision system is characterised by the decision-
making ability. As mentioned before in autonomouscontrolled production systems decision functions areshifted to logistic objects, which are aligned in aheterarchical organisational structure. These functionscontain planning and control tasks and enable logisticobjects to assign their progression. The decision-making
ARTICLE IN PRESS
So
Decision execution ability
SiSo
Resource
flexibility
Mobility
Output
Adjustment
Source/Sink
Machine
Material flow
Part / Lot
Break down
Conveyor
Data package
Database
Function
Identification
ability
Measuring
ability
M12
M12
M11 M21
M22
Information processing ability
Interaction
ability
Data processing
Date storage Type and location of
decision-making
Execution
instructionExecution
control
Legend
Decision-making ability
Planning
Time behaviour of
objective system
Objective system
Number of decision alternatives
Control
System
behaviourSelection of
alternative
Evaluation
of alternatives
Identification
of alternatives
Input
Execution system
Decision systemInformation system
Organisational structure
Fig. 1. System layers and criteria of autonomous control.
K. Windt et al. / Robotics and Computer-Integrated Manufacturing ] (]]]]) ]]]–]]] 3
Please cite this article as: Windt K, et al. Autonomy in production logistics: Identification, characterisation and application. Robot Comput Integr
Manuf (2007), doi:10.1016/j.rcim.2007.07.008
process includes the identification and evaluation ofdecision alternatives on the basis of an own, decentralisedobjective system, the selection, instruction and control ofthe best rated alternative as well as possibly adjustments.
The basis for decision-making is the informationprocessing ability on the information system layer. Inautonomous controlled production systems logistic objectsmust be able to interact with each other as well as to storeand to process data.
The execution system layer is characterised by thedecision execution ability of logistic objects. Autonomouslogistic objects are able to measure their current state andreact flexible to unforeseeable, dynamic influencing vari-ables. Mobility and high flexibility of the resources areother main criteria of autonomous control in productionsystems.
In the following a catalogue of criteria is derived, thatcontains the main criteria of autonomous control describedabove as well as its properties, which describe the differentlevels of autonomous control. The catalogue of criteria isillustrated in form of a morphologic scheme in Fig. 2.
The vision of autonomous control encloses transferringqualified capabilities (e.g. decision-making, data proces-sing, measuring) from the total system to the system
elements, i.e. autonomous logistic objects. So the visualisedsystem layers relate both to the total system and the systemelements. Each criterion has a series of properties, with anincreasing level of autonomous control in their order fromleft to right. For example, a logistic system withdecentralised decision-making by its elements has a higherlevel of autonomous control than a system renderingcentralised decisions. Grey marked properties show ex-emplary, how a considered production system could berepresented in the catalogue of criteria. Based on this, anexemplary scenario with the individual criteria and theirmarked properties are described in the following.
4. Application of the catalogue of criteria
In this Section, criteria and properties explained aboveare described using a production logistics scenario. Fig. 3gives an overview of a scenario of a two-stage job-shopproduction. Each criterion characterises the behaviour oflogistic objects and is assigned to different system layer, i.e.decision-making system, information system and executionsystem.The first production stage contains the manufacturing of
a part on two alternative machines (Mij). The raw materials
ARTICLE IN PRESS
System layer Criteria Properties
Decision
systemOrganisational
structurehierarchical
mostly
hierarchical
mostly
heterarchicalheterarchical
increasing level of autonomous control
Type of decision
makingstatic rule-based learning
Number of alter-
native decisionsnone some many unlimited
System behaviour
elements and
system
deterministic
elements and
system non-
deterministic
elements non-/
system
deterministic
system non-/
elements
deterministic
system layer subsystem layersystem-
element layer
Time behaviour of
objective systemstatic mostly static mostly dynamic dynamic
Information
system
Execution
system
Location of
data storagecentral mostly central
mostly
decentraliseddecentralised
Resource
flexibilityinflexible less flexible flexible highly flexible
Identification
ability
no elements
identifiable
some elements
identifiable
many elements
identifiable
all elements
identifiable
none data allocation communication coordinationInteraction ability
Measuring ability none others self self and others
Location of
data processingcentral mostly central
mostly
decentraliseddecentralised
Location of
decision making
Mobility immobile less mobile mobile highly mobile
System layer Criteria Properties
Decision
systemOrganisational
structurehierarchical
mostly
hierarchical
mostly
heterarchicalheterarchical
increasing level of autonomous control
Type of decision
makingstatic rule-based learning
Number of alter-
native decisionsnone some many unlimited
System behaviour
elements and
system
deterministic
elements and
system non-
deterministic
elements non-/
system
deterministic
system non-/
elements
deterministic
system layer subsystem layersystem-
element layer
Time behaviour of
objective systemstatic mostly static mostly dynamic dynamic
Information
system
Execution
system
Location of
data storagecentral mostly central
mostly
decentraliseddecentralised
Resource
flexibilityinflexible less flexible flexible highly flexible
Identification
ability
no elements
identifiable
some elements
identifiable
many elements
identifiable
all elements
identifiable
none data allocation communication coordinationInteraction ability
Measuring ability none others self self and others
Location of
data processingcentral mostly central
mostly
decentraliseddecentralised
Location of
decision making
Mobility immobile less mobile mobile highly mobile
Fig. 2. Extract of catalogue of criteria for autonomously controlled systems.
K. Windt et al. / Robotics and Computer-Integrated Manufacturing ] (]]]]) ]]]–]]]4
Please cite this article as: Windt K, et al. Autonomy in production logistics: Identification, characterisation and application. Robot Comput Integr
Manuf (2007), doi:10.1016/j.rcim.2007.07.008
that are needed for production are provided by the source(So). In the second production stage, the assembly of theparts that were produced in the first stage is donealternatively on two machines (Aij). The manufactureditems leave the material flow net at the sink (Si). At a pre-determined time a disturbance occurs in the form of abreakdown of machine A21. In conventionally controlledproduction systems a machine breakdown at night wouldcause at least a delay of many hours before recognising thedisturbance and adjusting the production plan in thetraditional way.
The autonomous control of the machines provides theopportunity to react fast and flexible to disturbances.Machine A21 recognises autonomously its breakdown bypermanent measuring and processing of sensors data.Deviations of sensors data are identified, analysed andappropriate activities are initiated. In this scenario, themachine A21 immediately informs other logistic objectsabout its breakdown, especially machine A22. Based on thisinformation, machine A22 could adapt its dynamic localobjective system by prioritising the objective high utilization
instead of low stock to counteract the bottleneck of theassembly stage. Parts waiting in front of machine A21 areinformed about the machine breakdown. Because of thisinformation and their measuring ability, parts can definetheir position and initiate their own transport to thealternative machine A22. Because of the identificationability, the conveyor is able to precisely identify the parts.
The existence of alternative manufacturing and assem-bling stages as well as the availability of local informationallows parts to render decisions regarding their waythrough the production process. The decision-makingprocess in this scenario is rule-based, i.e. logistic objectsact according to defined rules. For example, a part couldchoose the manufacturing machine on the basis of the rule‘‘select machine with lowest rate of utilization’’. Howeverin this scenario, parts are characterised by different levelsof autonomous control. Some parts just have the ability toallocate data; other parts acting for the entire lot are ableto navigate through the production process.On the basis of this exemplary scenario it has been
shown that each logistic system can be classified accordingto the level of autonomy by means of the introducedcatalogue of criteria.
5. Conclusions and further research
This paper introduced a definition of autonomouscooperating processes and its implementation in logisticsystems. Based on this definition, a catalogue of criteriawas developed and applied exemplary on a job-shopscenario. As shown in the scenario, establishing autono-mous control seems to be an appropriate method to copewith increasing complexity and dynamic environments oftoday’s logistic systems. In confirmation of this futureresearch has to deal with the development of a new
ARTICLE IN PRESS
Fig. 3. Autonomously controlled production logistics scenario.
K. Windt et al. / Robotics and Computer-Integrated Manufacturing ] (]]]]) ]]]–]]] 5
Please cite this article as: Windt K, et al. Autonomy in production logistics: Identification, characterisation and application. Robot Comput Integr
Manuf (2007), doi:10.1016/j.rcim.2007.07.008
evaluation system to estimate the achievement of logisticobjectives in autonomously controlled logistics systems.The evaluation system represents the degree of achievementof logistic objectives related to the level of autonomouscontrol. Therefore, both the degree of the achievement oflogistic objectives and the level of autonomous controlmust be measurable. Based on the catalogue of criteria thelevel of autonomous control of logistics systems can bedetermined with an adequate operationalisation. Further-more, the achievement of logistic objectives can beascertained through comparison of target and actuallogistic performance figures related to the objectives lowwork in process, high utilisation, low throughput time andhigh due-date punctuality. The evaluation system consistsof three evaluation steps to measure the logistic perfor-mance of the considered system. The first step evaluates thepossible decision alternatives, the second step the logisticperformance of individual logistic objects and the thirdstep the total system. Further research is directed towardsthe enhancement of the evaluation system to confirm thecoherence between achievement of logistic objectives andlevel of autonomous control in job-shop production shownin Fig. 4.
A low level of autonomous control in conventionalcontrolled logistics systems leads to a suboptimal achieve-ment of logistic objectives. An increase of the level ofautonomous control, e.g. by decentralisation of decision-making functions to the logistic objects causes a rise of theachievement of logistic objectives. However, at a certainlevel of autonomous control, a decrease of the achievement
of logistic objectives caused by chaotic system behavior canprobably be noticed. By dint of simulation studies theborders of autonomous control shall be detected to specifyin which cases an increase of autonomous control does leadto higher performance of the system any more.
Acknowledgements
This research is funded by the German ResearchFoundation (DFG) as the Collaborative Research Centre637 ‘‘Autonomous Cooperating Logistic Processes—A Paradigm Shift and its Limitations’’ (SFB 637).
References
[1] Scherer E. The reality of shop floor control—approaches to systems
innovation. In: Scherer E, editor. Shop floor control—a systems
perspective. Berlin: Springer; 1998.
[2] Scholz-Reiter B, Windt K, Kolditz J, Bose F, Hildebrandt T, Philipp
T, et al. New concepts of modelling and evaluating autonomous
logistic processes. In: Chryssolouris G, Mourtzis D, editors.
Manufacturing, modelling, management and control. Elsevier;
2006.
[3] Bose F, Lampe W. Adoption of RFID in logistics. In: Proceedings of
IBIMA international business information management association
conference, Cairo, 2005. p. 62–5.
[4] Freitag M, Scholz-Reiter B, Herzog O. Selbststeuerung logi-
stischer Prozesse—Ein Paradigmenwechsel und seine Grenzen. In:
Industrie Management, 20(2004)1. Berlin: GITO Verlag; 2004.
p. 23–7.
[5] Krallmann H. Systemanalyse in Unternehmen: partizipative
Vorgehensmodelle, objekt- und prozessorientierte Analysen, flexible
ARTICLE IN PRESS
Fig. 4. Achievement of logistic objectives vs. level of autonomous control.
K. Windt et al. / Robotics and Computer-Integrated Manufacturing ] (]]]]) ]]]–]]]6
Please cite this article as: Windt K, et al. Autonomy in production logistics: Identification, characterisation and application. Robot Comput Integr
Manuf (2007), doi:10.1016/j.rcim.2007.07.008
Organisationsarchitekturen. 3rd ed. Munchen, Wien: Oldenbourg;
2004.
[6] Scholz-Reiter B, Windt K, Freitag M. Autonomous logistic
processes—new demands and first approaches. In: Proceedings of
37th CIRP international seminar on manufacturing systems, Buda-
pest, Hungary, 2004.
[7] Hulsmann M, Windt K. Understanding autonomous cooperation
and control in logistics—the impact on management, information
and communication and material flow. Berlin, Heidelberg,
New York: Springer; 2007.
[8] Goldammer E. Heterarchie—Hierarchie: Zwei komplementare Bes-
chreibungskategorien. Download at 17.02.2006 from: /http://www.
vordenker.de/heterarchy/a_heterarchie.pdfS.
[9] Frese E, Schmidt G, Hahn D, Horvath P. Organisationsstrukturen und
Management. In: Eversheim W, Schuh G, editors. Betriebshutte,
Produktion und Management. 7th ed. Berlin, New York: Springer; 1996.
[10] Flamig M. Naturwissenschaftliche Weltbilder in Managementtheor-
ien. Frankfurt, New York: Campus Verlag; 1998.
[11] Ropohl GJB. Eine Systemtheorie der Technik—Grundlegung der
Allgemeinen Theorie. Munchen: Carl Hanser Verlag; 1979.
ARTICLE IN PRESSK. Windt et al. / Robotics and Computer-Integrated Manufacturing ] (]]]]) ]]]–]]] 7
Please cite this article as: Windt K, et al. Autonomy in production logistics: Identification, characterisation and application. Robot Comput Integr
Manuf (2007), doi:10.1016/j.rcim.2007.07.008