Professional Documents
Culture Documents
N 503788
N 503788
com
Available ScienceDirect
online atonline
Available www.sciencedirect.com
at www.sciencedirect.com
ScienceDirect
Procedia CIRP 00 (2018) 000–000
ScienceDirect www.elsevier.com/locate/procedia
ProcediaProcedia
CIRP 00CIRP
(2017)
72000–000
(2018) 1142–1147
www.elsevier.com/locate/procedia
Sensor-driven Analysis
28th CIRP Design of Manual
Conference, Assembly
May 2018, Systems
Nantes, France
A newKärcher*,
Susann methodology
Emir Cuk,toTimo
analyze the
Denner, functional
David and
Görzig, Lisa C. physical architecture
Günther, Anna of
Hansmersmann,
Günther Riexinger, Thomas Bauernhansl
existing products for an assembly oriented product family identification
Fraunhofer Institute for Manufacturing Engineering and Automation IPA, Nobelstrasse 12, 70569 Stuttgart, Germany
A reliable knowledge of processing times and following analysis are the basis for successful production planning. In particular, manual
assembly processes are associated with great efforts to get data about processing times.
Abstract
That is why manual assembly processes are often not sufficiently transparent to enable good production planning, scheduling and control.
This paper describes a system for automatic analysis and evaluation of manual assembly processes. It is driven by sensor and measurement
Intechnology,
today’s business environment,
constructed modularly, theand
trend
cantowards
be usedmore product
in a wide variety
range and customization
of assembly processes. is unbroken. Due to this development, the need of
agile and reconfigurable production systems emerged to cope with various products and product families. To design and optimize production
systems as well as to choose the optimal product matches, product analysis methods are needed. Indeed, most of the known methods aim to
© 2018 The Authors. Published by Elsevier B.V.
analyze a product or one product family on the physical level. Different product families, however, may differ largely in terms of the number and
Peer-review under responsibility of the scientific committee of the 51st CIRP Conference on Manufacturing Systems.
nature of components. This fact impedes an efficient comparison and choice of appropriate product family combinations for the production
system. A new
Keywords: methodology
assembly; is proposed
sensor; analysis; to modular;
manual; analyze evaluation
existing products in view of their functional and physical architecture. The aim is to cluster
these products in new assembly oriented product families for the optimization of existing assembly lines and the creation of future reconfigurable
assembly systems. Based on Datum Flow Chain, the physical structure of the products is analyzed. Functional subassemblies are identified, and
a functional analysis is performed. Moreover, a hybrid functional and physical architecture graph (HyFPAG) is the output which depicts the
similarity between product families by providing design support to both, production
1. Introduction In moresystem
than planners and product
sixty percent designers. Andata-driven
of companies illustrative
example of a nail-clipper is used to explain the proposed methodology. Andecisionsindustrial case
havestudy on or
a high twovery
product
highfamilies
priorityof[3].
steering
It is,columns of
therefore,
thyssenkrupp Presta France is then carried out to give a first industrial evaluation of the proposed approach.
Nowadays companies face the challenging environment of crucial to have reliable data. Assembly processes can only be
© 2017 The Authors. Published by Elsevier B.V.
an increasing number of product variants. Furthermore, the reliably planned, if reliable data are available.
Peer-review under responsibility of the scientific committee of the 28th CIRP Design Conference 2018.
trend towards decreasing lot sizes even down to one-piece In the field of assembly planning and optimization, the
production continues.
Keywords: Assembly; Design[1] In order
method; Familytoidentification
meet the challenges of need emerges to simplify the analysis of assembly processes
the changing environment and remain competitive, it is and the acquisition of data.
necessary to properly plan the production, especially the
assembly, in which seventy percent of production costs are 2. Objective
1.incurred
Introduction
[2]. However, there is often still too little of the product range and characteristics manufactured and/or
transparency within the system. Especially with manual assembled in this
This paper system.
aims In this
to present ancontext,
approach thetomain challenge
a flexible systemin
Due to processes,
assembly the fast itdevelopment
is difficult into the obtaindomain of
complete modelling
that enables andautomatic
analysis isanalysis
now notand onlyevaluation
to cope with single
of manual
communication
transparency due andto an ongoing
a lack trend of digitization
of standardization and
and detailed products,
assembly aprocesses.
limited product range or
The system existing
needs to beproduct families,
designed in its
digitalization, manufacturing
documented assembly enterprises
instructions. are facing
In general, important
the more time a but also to be
hardware, by able to analyze
sensors and to compare
and measurement productsastowell
technology, define
as
challenges in today’s
manual assembly processmarkettakes,environments:
the lower thea transparency
continuing new
in itsproduct families.
software. It can
The data be observed
create that classical
transparency existing
and are the basis
tendency
becomes.towards
But evenreduction of product
with shorter development
assembly processingtimestimes,
and product families are
for a successful regrouped
production in function
planning, of clientsand
scheduling or features.
control.
shortened
there oftenproduct lifecycles.
is no complete In addition, there is an increasing
transparency. However,
The system assembly
ensuresoriented
reliableproduct families
processing are and
times hardly to find.
allows an
demand
Due toof the
customization, being at the
missing transparency, it issame
oftentime in apossible
neither global On theofproduct
analysis family
the process level, products differ mainly in two
steps.
competition
to provide with competitors
specified all over
assembly the world.
processes norThis trend,
assembly main In characteristics:
the following the(i) the number
state of theofartcomponents
and research andgaps
(ii) the
are
which is inducing
instructions the development
in a production with smallfrom lot macro
sizestoormicro
long type of components
presented. (e.g.the
Afterwards, mechanical,
system is electrical,
introducedelectronical).
especially in
markets,
assemblyresults
processinsteps.
diminished lot sizes due to augmenting its Classical
setting, methodologies
IT-architecture considering mainlyFurthermore,
and evaluation. single products the
product varieties (high-volume to low-volume production) [1]. or solitary,
benefits already existing product families analyze the
are shown.
To cope with this augmenting variety as well as to be able to product structure on a physical level (components level) which
identify possible optimization potentials in the existing causes difficulties regarding an efficient definition and
2212-8271 ©system,
production 2018 The it
Authors. Publishedtobyhave
is important Elsevier B.V. knowledge
a precise comparison of different product families. Addressing this
Peer-review under responsibility of the scientific committee of the 51st CIRP Conference on Manufacturing Systems.
2212-8271©©2017
2212-8271 2018The
The Authors.
Authors. Published
Published by Elsevier
by Elsevier B.V. B.V.
Peer-review
Peer-review under
under responsibility
responsibility of scientific
of the the scientific committee
committee of the of theCIRP
28th 51stDesign
CIRP Conference
Conference2018.
on Manufacturing Systems.
10.1016/j.procir.2018.03.241
Susann Kärcher et al. / Procedia CIRP 72 (2018) 1142–1147 1143
2 Author name / Procedia CIRP 00 (2018) 000–000
the context of manufacturability of the product and segmentation and encoding the data into continuous strings
accessibility of the mounting location. An assembly planning which are matched against motion string templates [11]. The
system with motion capture and VR techniques is already approach reached activity recognition rates close to 100
presented by Bullinger et al. [19]. Electromagnetic sensors percent in the first scenario and 74 percent in the second
track positions of the head, hand and feet for a virtual setting. In [23] a single wrist-worn IMU is sufficient to
representation of the operator’s body. Data gloves capture distinguish between hammering, screwing with a screwdriver
information on the orientation of the hand and fingers to or a power drill, spanner use and null data with an average
analyze the interaction between the operator and virtual accuracy of almost 88 percent. The classification was done
objects. Gestures detected during the virtual assembly and the using features derived from acceleration and angular speed
disassembly of components are set into a relationship to MTM data, as well as input into a nearest neighbor method.
basic movements to calculate planning times for the assembly In contrast to the human-centric motion capture approaches
operation and a logical assembly sequence. A similar [24] focuses on real-time object tracking in a manufacturing
approach is carried out by Ma et al. [20]. A hybrid tracking environment. Wang et al. propose to attach RFID tags to
system, including eight CCD cameras for optical motion- every object moving along an assembly line and to evenly
tracking and a data glove, collects information on body equip the working area with RFID readers. This allows for
motion and hand gestures of a worker performing a manual real-time object position and movement path prediction and
handling task. After performing automatic motion tracking, thereby permitting real-time planning and control of
recognition, the operation is evaluated with the Maynard the manufacturing assembly line.
Operation Sequence Technique (MOST), a simplified motion- As a result of industry 4.0, the number of assistance
time evaluation derived from MTM. Both approaches show systems for biomechanical analyses has increased. Assistance
limitations: [19] is limited to seated operations (see systems can be applied to provide process workflow and
[20] p.341); in [20] not all MOST motions could be assembly support for the operative personnel, using computer
recognized automatically. Instead of realizing a fully vision-based tracking systems such as depth cameras to
automated time study Elneklave and Gilad [21] improve the capture e.g. the withdrawal of material or to recognize parts
traditional videotape analysis approach. On average, time and their orientation. [25] Parallel in-situ projections or
savings of 43 percent were attained by using enhanced video visualizations of workflow and process data with mixed reality
software during the conduction of MOST analysis for five systems provide additional support for the targeted and guided
case studies, compared to a standard video cassette recorder. production process. Component mix-up can be prevented and
This software includes advanced playback options, the assembly process is less error-prone [26].
measurement of distances implicated in the tasks executed by
the operators and a work sampling module. 4. Research gaps
Among other approaches, motion capture is also applied to
the training of manufacturing personnel. Müller et al. The state of the art shows that systems for the detection of
supplement a workplace with a Kinect™ sensor and a manual assembly steps already exist. These systems are
learning environment containing computer supported generally used for quality monitoring or training, for example
instructions and an e-learning module. An algorithm extracts to recognize whether all the necessary process steps have been
the position of the worker’s hands from the sensor data, taken. The systems normally must be taught in advance: It is
determines the individual operation and informs the operator usually necessary to define a selection of possible processes
about the next step of the assembly process. During the study, or acquire an exact knowledge of the processes. But in
the initial learning of the given assembly process could not be particular, when assembling with long processing times, there
accelerated compared to the training with paper-based is no predefined process order or information about individual
instruction due to – among other things – challenges in hand process steps are missing. Therefore, a more flexible system is
tracking in different lighting conditions [22]. Stiefmeier et al. necessary. Other limitations of previous approaches using
[11] choose wearable activity tracking to support the training sensors to record manual assembly processes are amongst
and work of assembly line operators. The study emphasizes others:
the importance of redundant and diverse sensor systems to • Most systems use body-worn sensors: The human-oriented
ensure reliable tracking of assembly steps. Apart from evaluation of work processes is one approach to acquire
wearable sensors (Inertial Measurement Unit (IMU), Radio processing times. This could cause problems in the field of
Frequency Identification (RFID) reader, force-sensitive privacy protection and these sensors can handicap the
resistors (FSR)) further sensors are placed on tools (IMU, worker.
RFID tags) and the main assembly (FSR, magnetic switches) • The use of systems with motion tracking is often
where the sensors on the assembly were most crucial for associated with great effort (costs, time). Moreover, the
activity recognition. The task tracking is implemented through movements of humans are recorded in great detail. In
with a finite state machine approach that identifies the terms of production optimization, however, the process
transition between individual activities by the presence and view is more interesting than human movements.
absence of signal combinations. In the second scenario, which • The assembly detection systems are so far essentially
was not conducted in a training environment, 27 sensors were aligned and evaluated in test environments (no real
integrated into a jacket, as sensors on assemblies were not workspace environment, short duration).
applicable. Activity recognition was done through data
Susann Kärcher et al. / Procedia CIRP 72 (2018) 1142–1147 1145
4 Author name / Procedia CIRP 00 (2018) 000–000
5. Analysis system
Sensors and measurement technologies are installed on The system is built in Node.js and is responsible for the
parts, products, tools and operation materials. There are no data acquisition, device management, processing of the data
sensors installed on workers. The system uses contacting and and the understandable visualization of the results as well as
non-contacting sensors and the recording is wireless. the adjustment of the data processing. The goal is to provide
To use the system, the sensors have to be positioned and the user with high flexibility and high fault tolerance. The
installed. After this, no further intervention by an employee is presented system uses MongoDB, a NoSQL database.
necessary and the data acquisition runs. The following figure It obtains all information through external sensors. These
shows tools equipped with sensors. sensors are always a part of a device by definition. Only
devices are able to communicate directly to the system, in the
present case via Wi-Fi and USB. A sketch of the device and
the sensors connected to the system is visualized in Fig. 4.
1146 Susann Kärcher et al. / Procedia CIRP 72 (2018) 1142–1147
Author name / Procedia CIRP 00 (2018) 000–000 5
As soon as the system has recognized which process steps [2] Gairola A. Montagegerechtes Konstruieren. Ein Beitrag zur
are carried out, optimization methods can work on the data Konstruktionsmethodik. Diss. TH Darmstadt; 1981. p. 9.
[3] Bauer D, Maurer T, Henkel C, Bildstein A. Big-Data-Analytik:
basis. Datenbasierte Optimierung Produzierender Unternehmen. Studie
Stuttgart; 2017. p. 20.
6. Benefit [4] Lotter B, Wiendahl H-P. Montage in der industriellen Produktion. 2.
Auflage, Berlin Heidelberg: Springer Verlag; 2012.
The approach presented in this publication provides great [5] REFA Methodenlehre der Betriebsorganisation – Grundlagen der
Arbeitsgestaltung. München: REFA Bundesverband e. V.; 1993. p. 81-
benefit: Manual assembly processes can be evaluated and the 125.
efficiency and effectiveness of the assembly process can be [6] TiCon TS, in: Apple App Store https://itunes.apple.com/de/app/ticon-
improved. The system creates transparency, so that manual ts/id1207860674?mt=8 last accessed 30-11-2017
assembly processes can be planned, scheduled and controlled [7] UMT Plus, in: Apple App Store https://itunes.apple.com/us/app/umt-
better. This will save costs and time along with an plus/id501809425?mt=8 last accessed 30-11-2017
[8] W2MO Process Study, in: Google Play Store
improvement in quality. https://play.google.com/store/apps/details?id=com.logivations.w2mo.mob
Improved planning leads to superior planning of capacities ile.time.study last accessed 30-11-2017
as there is now transparency on the time individual process [9] Picker C. Prospektive Zeitbestimmung für nicht wertschöpfende
steps take. Efficiency will increase and assembly processes Montagetätigkeiten. Technische Universität Dortmund, Fakultät
can be improved and optimized. In addition, orders can be Maschinenbau, Dissertation. Aachen: Shaker Verlag; 2006.
[10] Syska A. Produktionsmanagement. Wiesbaden: Betriebswirtschaftlicher
planned better with reliable information on throughput times. Verlag Gabler; 2006.
The system also provides an essential advantage in the [11] Stiefmeier T, et al. Wearable activity tracking in car manufacturing.
flexible analysis: It can be applied in all use cases and high IEEE Pervasive Computing, 2008:7(2):49.
time savings for analysis and further evaluation can be [12] Gudehus T. Entwicklung eines Verfahrens zur ergonomischen
achieved. Moreover, the historical evaluation can be changed Bewertung von Montagetätigkeiten durch motion-capturing. Kassel:
kassel university press GmbH; 2009.
by adjusting the limit values and tolerances later on. [13] Karhu O, Kansi P, Kuorinka I. Correcting working postures in industry:
Manual assembly processes are automatically and a practical method for analysis. Applied ergonomics, 2006:8(4):199-201.
continuously analyzed. This leads to a homogeneous result of [14] Bellmann VK, et al. Development of a System for a Real Time
all considered processes. Even if several processes are Ergonomic Assessment. 3rd International Conference on Production
considered on different days, the analysis is always the same. Automation and Mechanical Engineering on 26th - 27th August 2017, in
Montreal, Canada, 2017.
In the long term, a database will be set up with experience [15] Härtel T, et al. Capturing and Assessment of Human Motion during
data which could then be used for other assembly tasks. Manual Assembly Operation. First International Symposium on Digital
Additionally, the system does not interfere with the process Human Modeling, Conference Proceedings, 2011.
and the process does not have to be interrupted. Further [16] Diego-Mas JA, Alcaide-Marzal J. Using Kinect™ sensor in
advantages are the choice of sensors, because no sensors are observational methods for assessing postures at work. Applied
ergonomics, 2014:45(4):976-985.
attached to workers and no optical motion detection takes [17] Mgbemena CE, et al. Gesture Detection Towards Real-Time Ergonomic
place. Analysis for Intelligent Automation Assistance. Advances in Ergonomics
of Manufacturing: Managing the Enterprise of the Future. Basel: Springer
7. Conclusion and outlook International Publishing; 2016. p. 217-228.
[18] Schirra R. Entwicklung eines Systems zur immersiven
Arbeitssystemgestaltung mit CAD-Schnittstelle. Diss. Uni Stuttgart;
In conclusion, an approach for automated capture and 2009.
analysis of manual assembly operations has been presented. [19] Bullinger HJ; Richter M, Seidel K‐A. Virtual assembly planning. Human
Manual assembly processes can be recorded and evaluated, Factors and Ergonomics in Manufacturing & Service Industries,
saving time and creating new possibilities for evaluation. 2000:10(3):331-341.
Further work will be carried out to improve accuracy and [20] Ma L, et al. A framework for interactive work design based on motion
tracking, simulation, and analysis. Human Factors and Ergonomics in
to transfer the prototype to industrial application. With a Manufacturing & Service Industries, 2010:20(4):339-352.
rising number of implemented use cases the data base is [21] Elnekave M, Gilad I. Rapid video-based analysis system for advanced
constantly enhanced and due to additional sensors, more and work measurement. International journal of production research,
more data will be collected. In addition, the evaluation 2006:44(2):271-290.
algorithms are improving. Consequently, the evaluation will [22] Müller BC, Nguyen TD, Dang QV, Duc BM, Seliger G, Krüger J,
Kohl H. Motion Tracking Applied in Assembly for Worker Training in
be more accurate. In the future, less information will have to different Locations. Procedia CIRP, 2016:48:460-465. p. 461.
be made available to the system. [23] Koskimaki H, et al. Activity recognition using a wrist-worn inertial
Furthermore, the negative impact of disturbing influences measurement unit: A case study for industrial assembly lines. Control and
should be investigated. Also, the variable granularity or Automation, 2009. MED'09. 17th Mediterranean Conference on. IEEE;
detailing can still be adjusted; however, more experience in 2009:401-405.
[24] Wang J; Luo Z; Wong EC. RFID-enabled tracking in flexible assembly
industrial application is necessary. Lastly, the influence on the line. The International Journal of Advanced Manufacturing Technology,
worker by the system is to be analyzed. 2010:46(1):351-360.
[25] Krüger J, Wang L, Verl A, Bauernhansl T, et al. Innovative control of
References assembly systems and lines. CIRP Annals, 2017:66:707–730.
[26] Fraunhofer IPA. https://www.ipa.fraunhofer.de/de/presse/presseinforma
[1] Koren, Y. The Global Manufacturing Revolution. Hoboken, New Jersey: tionen/2015-05-18_gruenes-licht-fuer-naechsten-montageschritt.html
last accessed 30-11-2017
John Wiley & Sons, Inc; 2010. p. 15.