An IoT Framework for Assembly Tracking and Scheduling in

Manufacturing SME

Meysam Minoufekr

, Anass Driate and Peter Plapper

University of Luxembourg, 6 rue Coudenhove-Kalergi, Luxembourg

Keywords: Industrial Internet of Things, IIoT, Internet of Things, IoT, Internet of Production, Assembly Tracking,

Industry 4.0, Production Management.

Abstract: A universal RFID platform is presented, which is meant to be used as a building component of IoT integrated

collaboration platform for manufacturing applications. The core element of the system is based on affordable

Raspberry Pi modules, running on an IoT operating system. The main goal of this paper is to demonstrate an

affordable IoT solution for manufacturing SME to improve productivity by measuring and adapting the

assembly processes for a given product. To track the production chain, each part in the supply chain is

equipped with an RFID tag, which will be recorded during its travel through the facility. In addition, each

worker has his own RFID tag to localize himself and record the performed activities. The workstations are

equipped with RFID scanners used to record activity and product flow through the stations. All the gathered

data is collected on a server and the real-time status of the assembly line is processed and displayed to the

dispatching agents. Upon this data analysis, the dispatchers can take actions, update the manufacturing setup

and ultimately increase productivity.

1 INTRODUCTION

The Industrial Internet of Things (IoT) is a concept in

which people are surrounded by objects that are all

interconnected using embedded technologies – smart

objects (Kortuem, 2010). In the IoT, these devices can

exchange information and process it via network,

providing a next generation of services to humanity.

The environment is packed with systems that play the

role of sensors and actuators (Gubbi, 2013). All these

devices are spatially distributed, and communication

between them may be established using a variety of

technologies (

Ebling

, 2017).

These technologies have to be integrated with

easiness of creating ad-hoc connections for seamless

development and operation of the Internet of Things

(Ray, 2018). Currently, most implementations of IoT-

like solutions are being developed for business

solutions affordable for large enterprises

(Uckelmann, 2016), but there is an emerging need for

extending it to the small and medium size enterprises

(SME), with standardized and easy- to-use solutions.

The Scope of this paper is to make use of RFID

technology to record live data of assembly lines,

https://orcid.org/0000-0002-5877-0820

process this data and visualize it to facilitate decision-

making, layout and scheduling optimization for the

given assembly line. This will involve tracking the

parts and workers of the facility, introducing a digital

real-time interface for workstations and visualization

of performance statistics for dispatching personnel

optimizing the assembly processes.

The structure of this paper is as follows: the first

step is to understand the challenges of manufacturing

small and medium sized enterprises (SME). The

second step is a to analyse the existing technological

boundaries and introduce the RFID tracking to

achieve an implementation strategy followed by the

actual solution, which represents the development of

the measurement and workstation unit and its

deployment on the local assembly-line or on a smaller

test assembly-line. Ultimately, the RFID-based

system within the existing assembly line and provide

new ways to visualize the assembly process, its

bottlenecks and in the end provide useful statistics

and analytics.

Minoufekr, M., Driate, A. and Plapper, P.

An IoT Framework for Assembly Tracking and Scheduling in Manufacturing SME.

DOI: 10.5220/0007921805850594

In Proceedings of the 16th International Conference on Informatics in Control, Automation and Robotics (ICINCO 2019), pages 585-594

ISBN: 978-989-758-380-3

585

1.1 Background and State of the Art

Internet of Thing (IoT), concept of physical objects

exchanging data through the Internet, is entering

enterprise market bringing growth and profit. In order

to function properly, IoT comprises of many different

abstraction layers from tiny sensors detecting

temperature change in individual objects up to huge

server warehouses powering cloud- computing

technologies. Before data coming from millions of

‘real’ physical objects can be processed, these objects

have to be somehow mapped into virtual world. This

is done by using sensor technologies that are able to

transform physical properties into electric signals.

One of the sensors, regarded as fundamental in IoT

world is Radio Frequency Identification (RFID)

technology. The RFID technology offers the ability to

detect and distinguish individual objects from

distance.

A big advantage of RFID is that it is affordable

for any manufacturing SME by using modern RFID

systems, which are based on commonly available

technology (Bucciero, 2018). The approach presented

here, can easily be adapted to new fields of

application, such as pharma production or healthcare

(Miorandi, 2012). Manufacturing companies are

strongly moving away from a push- to a pull

manufacturing, which requires highly flexible on

demand assembly lines. Such lines feature the

challenge of stock management and just-in-time

supply management for the assembly lines. To stay

competitive, manufacturing SME must produce with

high quality standards and avoid scrap parts. The cost

of reassembling, reworking products might quickly

become higher than the cost of the product itself and

result in direct waste.

The main manufacturing tasks addressed in this

paper are manual assembly processes. The assembly

can vary strongly between applications and product to

be manufactured. In general, the worker will

manufacture the product by following a certain

amount instruction, which guide him/her through the

assembly process on the workstation. Those

instructions are printed on paper separated into

smaller steps. These steps are to be executed within a

given time frame and with a defined quality that

addresses mostly human errors. An assembly error

can result in a rework or scrap product-creating waste

of time and resources. Tracking the assembly steps

enables crosschecks at each step and increases the

chance that products are assembled correctly,

reducing rework and production time.

2 PROBLEM DEFINITION AND

SOLUTION

Before data coming from numerous ‘real’ physical

objects can be processed for assembly tracking, these

objects have to be somehow mapped into virtual

world. This can be done by using sensor technologies

that are able to transform physical properties into

electric signals. One of the sensors, regarded as

fundamental in IoT world is Radio Frequency

Identification (RFID) technology. The RFID

technology offers the ability to detect and distinguish

individual objects from distance.

Current RFID systems are usually designed to

gather data about recognized objects and to send them

right away to a computer for processing. This solution

might be enough for smaller number of RFIDs (e.g.

timing marathon and races) but will be overwhelming

for the computational unit in IoT domain with

increasing investment and maintenance costs to the

manufacturing company.

This problem can be solved by adding smaller

computing units to each RFID reader. This computing

unit would process the data on the spot and send

results only when it is desired. This paper will

concentrate on creating such a platform (computing

unit with RFID reader, later called RFID platform) by

assembling hardware parts and writing necessary

software.

Figure 1: Requirements analysis addressing the challenges

to the system.

In this paper, we present a solution based on low

cost electronic tags that can easily be attached to

physical objects, and do not require power source –

reader in another device (Quan, 2018) initializes

them. Emerging from the determined Factors, the

ICINCO 2019 - 16th International Conference on Informatics in Control, Automation and Robotics

586

requirements to the system are formulated in Figure

1. The different requirements are explained as

follows:

Identify Workers (R.1): Each worker must be

uniquely identifiable by the system. This will assist

with building statistics over user productivity.

Knowing the real-time activity and status of each

worker is viable for tracking an assembly line. There

are a multitude of ways to identify Humans which

range from cheap and quick like a “user-password”-

combination to highly secure and customizable

solutions such as an iris-scanner. The most suitable

option for identifying workers at their workstations is

the RFID-Wristband which can be easily scanned

contactless and with no further doing from the

worker. The installation of a RFID sensor at each

workstation is at a low cost of less than 10 Euros and

directly compatible with the data-collection

technologies of choice.

Identify Objects (R.2): Identifying objects means

in this context to be able to identify parts, products,

tools and other hardware required in the

manufacturing process. These objects can differ

greatly in shape, size and portability. It is important

to therefore find reproducible and flexible ways to

recognize objects throughout the assembly line.

Recognizing objects cannot be done without further

doing. Objects are more versatile in shape

combinations than humans are. This eliminates

options such as fingerprint scanners and leaves us

mostly with option, which require the object to be

marked.

Communicate between Nodes (R.6): Different

nodes of the system need to be able to send data and

triggers over a network to be defined. The less the

latency and complexity of this network, the better the

results.

Collect and Store Data (R4): In order to

electronically process data, it first needs to be

acquired and stored in the system. This process

should be automatized and cost the least amount of

time possible. Each fluctuation in the acquiring of

timestamps will bias the dataset. The Raspberry Pi 3

device offers many possibilities, which are partly

offered by other platforms too but the small form,

factor clearly beats the other options when comparing

compatibility and functionality/size.

Measure Worktime (R.3): The time a worker

spends on a product, a task or which an object in

question is dependent on factors, which cannot

directly be known. It is vital tough to track the

assembly speed for each task and build statistics,

which enable to better understand the challenges of

each assembly line. The events registered when an

RFID-tag is swiped will determine the time-window

used for each operation. For operations, which do not

incorporate any tagged object or product, a physical

button will be introduced within the Users workspace,

which he/she can operate to signalize the start and end

of a timed event.

Process the Collected Data (R.5): The data, which

is stored in the system, needs to be processed in real-

time to be visualized for the assembly-line

supervisors in time. This is taken over by an

electronic component handling all the calculations of

the system.

Visualize Results (R.7): The computed results

finally must be transformed into human

understandable and readable output. There are

different media or transferring information to a

worker. This depends on the complexity of the data

and the capability of the worker. The latter is

influenced by how busy the worker is or how

susceptible he/she is to notice a change in

information.

Analyze and Improve (R.8): Collecting and

visualizing data does not mean the improve of the

productivity. This stage is where you turn your

measurements into insights and actions. Analyzing

the data and knowing what is wrong and the parts of

the process that need fixing, offer you the time to put

improvements in place and get them stick. This is the

best way for continuous improvement.

3 IMPLEMENTATION

This section shows how IoT can be realized to

improve production efficiency by proposing an

approach to support IoT-based assembly processes

through integrating manufacturing data. In order to

test the proposed approach and assess its impact on

improving production efficiency, a pilot study was

carried out in a discrete assembly line in our

university’s Lean Lab. Several smart devices with

sensor have been installed at assembly station level to

collect production data in real-time, and then this data

have been analyzed to see improvements of

production efficiency. The IoT solution platform

proposed in this paper, will address the challenges

An IoT Framework for Assembly Tracking and Scheduling in Manufacturing SME

587

Urgency, Complexity and Security by providing an

approach which

 Can be easily introduced in existing production

line (plug-n-play), without disturbing and

interrupting the production capacities,

 Is easy to use and based on standard components

and thus affordable for producing SMEs and working

with low configuration efforts, hiding technological

details from the users

 Is using standardized security protocols and

assures (at least) the same safety measure to protect

user and customer data like the one given by the

company’s IT infrastructure

The lack of understanding of manual assembly

process behavior “who”, “where”, “when”, and

“how” is the essential reason of the difficulty in

evaluating and improving production efficiency. In

order to improve production efficiency, significant

attention and efforts have to be made to obtain

process data from smart devices, sensors, and other

tools (Saeid, 2017), and then integrate this data in

production management. As a result, the approach to

build a model for supporting IoT-based production

management in sustainable smart factories can be

summarised as follows:

1. Understanding assembly processes and current

manufacturing management practices

2. Monitoring and analysing manufacturing

progress in real-time using IoT technology (e.g. smart

sensors)

3. Integrate process data into factory’s tools for

improving production efficiency

4. Define sustainable strategies and practices in

production management to improve production

efficiency

The first phase involves the understanding of the

production processes, the evaluation of the current

production management practices and the definition

of key performance indicators (KPIs). The second

phase focuses on collecting (possibly) real time data

by means of IoT technology and then analysing them

to understand current practices and limitations. Here

it is required to define and model the assembly

stations to be monitored, define the monitoring

devices for each station, the communication system

and where and how the data will be stored and

analysed. Moreover, the production processes must

be identified (e.g. assembly sequence, processing

time for each product under different configurations),

so as to link and understand the process behaviour and

make the efficient decision. After collecting and

analysing the data, the third phase is to integrate this

data into production management tools to enable the

decision makers to define the waste, where

improvement can be achieved, also select the most

sustainable configuration mode of assembly setup

with considering the production planning. The fourth

phase encompasses the upper level, i.e. the definition

of strategies and practices to improve the production

efficiency of the smart factory “by design”, for

example by integrating production data in production

management practices.

The IoT platform proposed in this paper majorly

consists of four main components and is based on the

architecture proposed by Kranz (Kranz, 2017):

Sensors / Actuators: RFID Sensors collect data from

parts and workers in each assembly station and

measure useful data like process time. Actuators

intervene to change the process flow, for example by

alerting the worker at the assembly station in case of

process errors.

Edge Devices: The device systems build the control

element of the sensors and feed data for data

aggregation on the server. In our case, Raspberry PI3

are used as edge devices, which are physically

attached to the RFID sensors and logically connected

to adjacent gateway device or server for data

forwarding and processing.

Gateways: A gateway provides connectivity between

all elements in the physical world and digitalized

cloud data. In our use case application, we use a

standard Wi-Fi gateway switch.

Cloud/Server: Although edge IT processing systems

may be in remote locations, our server backend

application is in our Lean Lab facility.

The implementation is focussed on the assembly

process of a paper puncher tool. The according

assembly line is composed of seven stations: chassis

assembly – base assembly – marriage – QA &

shipping – receiving product and dismantling 1 –

dismantling 2 – material handling. Each station

consists of many individual sub-assembly tasks. The

last station (Material handling) involves motion,

storage, and delivery of materials throughout the

process.

ICINCO 2019 - 16th International Conference on Informatics in Control, Automation and Robotics

588

In the proposed IoT framework for puncher

assembly processes, the overall objective is to

complete a well-defined life cycle, which will result

in the assembly of a target paper puncher device. To

accomplish this objective, a set of distributed IoT

components and devices including cyber physical

resources and modules collaborate. A well-defined

set of functional activities are accomplished, and they

include the following:

Creating the assembly plan

Creating of the 3D assembly instructions

Analysis of assembly alternatives using prediction

models to assess the assembly

Physical assembly of target parts using IoT

These resources are linked via the Internet using

a cloud computing approach. The components of this

cyber physical framework for micro assembly are

shown in Figure 2.

Figure 2: Overall system design and approach.

In our implementation of our IoT framework and

the associated remote collaboration protocols, a

remote client-server approach is realized that is based

on standard Windows setup as a server and Windows

IoT platform (Sabanal, 2016) Provisioned on bare

Raspberry Pi3 hardware for the assembly node

clients.

This application hosts the assembly planning and

real-time tracking (created using C#, Windows IIS)

that would be the target for remote clients interested

in the collaborative analysis in our IoT framework.

Remote users with thin clients are able to connect to

the server applications. Users from different

geographical locations are able to collaborate using

thin clients to propose assembly plans and study

assembly plans.

We will now present implementation of the

previously discussed aspects. The setup is shown in

Figure 2. Each station consists of a combination of

one IoT device (Raspberry Pi3) combined with an

RFID sensor, which is assembled on a PCB, attached

to the top part of the 3D printed frame holding the IoT

device. Each IoT device is equipped with a touch

display to interact with the worker. The

communication between the Pi and the sensor via the

SPI protocol (Spi et al., n.d.).

The basic workflow of one station is as follows:

after an initialization phase, a worker can login and

then start the execution of a task (Figure 3). The

worker moves from one-step of a task to the next,

using one of several methods, as discussed later. At

some points in the workflow, there are interactions

with the server, specifically for validation before

proceeding. The instructions are displayed to the

worker with both text and images, to facilitate the

understanding of the tasks and to ensure a smooth and

productive operation. The application was developed

according to the following structure chart, in which

all basic activities and their correlations are shown,

see Figure 3. When a RFID tag is detected at the

sensor, a request is sent to the server to identify to

which group it belongs (workers, parts, etc.). After

receiving the answer, the station sends a login

request. Finally, it proceeds when receiving a positive

answer.

Figure 3: Sequence for logging in a worker.

The main menu consists of four buttons, which

can be used to place orders, register, search for

products and navigate through the application. In the

activity “new job”, two buttons can be used to decide

for which factory the order is to be created. When the

activity is started, an RFID tag can be scanned. The

RFID code displayed on the contains the component’s

data and the according instructions, which are

displayed on the screen.

Figure 4: Structure of the application.

The second important part of the implementation

is the centralized data communication, see Figure 4.

It is the central element of the assembly network,

interacting with all involved devices, be it for the

storage, validation or retrieval of data. The secure

data communication between the app and the

database is realized by a web server regulating the

An IoT Framework for Assembly Tracking and Scheduling in Manufacturing SME

589

communication between the app and the database and

making the data available directly in a suitable form.

The production data in the production system are

stored in an SQL database in tables, which are divided

into customer data and order data. Both the database

and the web server are located on one server, running

independently. Only a single interface enables

communication between the web server and the

database.

The Database access via the web server takes place in

four steps:

1. The app sends an http request to the web server.

The http request contains the information what is

requested (order, customer, events). Furthermore, the

http request contains attributes that are necessary for

the web server to search (for example, an order

number when requesting an order).

2. The web server checks the syntax of the request.

If the syntax is correct, the web server queries the

desired data in the database using an SQL statement.

3. If there is an entry for the attribute passed (such

as the order number), the web server gets the data

back and creates an xml document. If there is no entry

for the attribute, an empty document will be created.

4. The web server responds to the http request with

the created document containing the desired

information. These are encrypted for security reasons.

Figure 5: Database table representing a worker.

Objects are identified by RFID stickers (R2), see

Figure 6. Another aspect to be considered in this

context is the detection range of RFID tags. The latter

increases both with the size of the RFID tag and

sensor and with the power of the sensor. A trade-off

between shape, size and power has therefore to be

found, tailored to the specific situation on the shop

floor. It is therefore important that the system can be

adapted to different configurations and types of

sensors.

The components are represented using an SQL

database. The basic structure of the tables involved is

shown in Figure 5. Moreover, each performed step is

associated with an entry in the AssemblyInstance

table. The main information are the start and the end

time of the step. Furthermore, each workstation is

associated to the instructions performed on it, the

workers on that workstation and the scanned objects

(StationFlow table). The latter allows regrouping all

the steps for one specific instance of a product.

Figure 6: Tagged object and several tag alternatives.

The sequence of steps when scanning an object is

like the identification of a worker, with first the

identification of the “owner” of the tag, before

sending a request to the server. When validated, the

station shows the next step of the task. If at any point

in time there is an error, the user receives a visual

feedback, as illustrated in Figure 7. The screen flashes

green in the case of a success.

Figure 7: Visual feedback for user if scanning a part was a

success or an error.

The measurement of the time a worker spends on

a given task is measured for each step. To indicate

that a step is finished, the worker either presses a

button or scans an RFID tag. In both cases, the station

directly sends a request for the validation to the

server. When the server acknowledges the completion

of the step, the station goes on to the next step or starts

ICINCO 2019 - 16th International Conference on Informatics in Control, Automation and Robotics

590

Figure 8: Database tables for saving process data.

An example of recorded data is shown in Figure

9. It is important to note that this is only the raw data

that must be processed before using it to analyze the

process performance. This basic data can be used to

obtain a lot of different information, a simple example

being the cycle time or identifying the best/worst

station.

The processing and visualization of the recorded

data is highly interlinked. Information is extracted out

of the data at the moment it is requested, in order to

always have the newest information available. The

chosen technology for displaying results is an HTML

page, as this makes it easily deployable across

different platforms and media.

Figure 9: Example process data.

When the IIS server receives a request, as

illustrated in Figure 10, a controller first extracts the

data from the database, before passing it on to a view

to setup the page. This is done via data objects that

are independent of the exact visualization of the data.

This allows for an independent development and

testing of both components as well as an easy

exchange of just one part of the program. After

receiving the data, the view then transforms the data

into the exact form it needs it to be, before generating

the whole page with the aid of the Razor technology.

Finally, the page is being returned to the client.

Currently two main pages are in use. Both display

their information per station. The first one presents

some basic statistical information about the process

performance. It includes the average cycle time, the

last cycle time, the number of products finished as

Well as an indication in which state the station is in

(e.g. operational or error). Furthermore, it displays

the stations with the minimum and maximum on the

cycle time and the number of products finished. The

second page displays the times of the different steps

for each station via a Gantt chart. It includes three

bars per step, indicating the minimum, average and

maximum duration of that step. This is a useful tool

to evaluate which steps for one specific station take

the longest, whereas the first page is more useful to

compare between stations. Both pages can be seen on

figures in the next section Discussion and Result.

As previously discussed, the connection method

between the server and the stations is done via WiFi,

which allows for a flexible rearrangement of the

stations, without needing to pay attention to the extra

effort of placing network cables. The topology of the

network is a star-based, with the server in the centre

of a “cloud” of stations. The client displaying the

information pages can be any device with a browser.

It is important though that the display attached to it is

big enough for proper of rendering of the information.

Figure 10: Structure of a request to display live data.

4 RESULTS AND DISCUSSION

The “Lean Assembly Lab” at the University of

Luxembourg is a testbed to practice concepts of Lean

manufacturing and basic concepts of Kaizen. Both

concepts are state-of-the-art methods in industrial

application to help increase efficiency in production.

In this section, we present some results, which shows

the advantages, in terms of efficiency, productivity

and scalability, of using the proposed IOT platform.

Two experiments have been carried out at the lean

assembly lab. Both tests have been performed in the

same environment: process of hole punching tool, six

stations, one worker per station, same process layout,

and same product (Table 1).

Those experiments are aimed to distinguish

between effects resulting from the paper-based

An IoT Framework for Assembly Tracking and Scheduling in Manufacturing SME

591

assembly process and those resulting from a more

comprehensive, systemically transformation by

digitalization of an Assembly process.

The key question we investigate in this section is:

“Given a lean process, what is the impact of applying

the described IoT framework to that process

regarding process improvement?”

The Layout design is an important component of

an assembly line and other operations, both in terms

of maximizing the effectiveness of the production

process and meeting the needs of employees. The

basic objective of layout is to ensure a smooth flow

of work, material, and information through a system.

Table 1: Conditions of the experiments.

Condition Value Units Symbol

Net time

available to

work

60 min

T

Customer

demand

20 units D

Number of

worker / station

1 NA NA

Number of

stations

6 NA NA

One worker is required on each station to

complete the job and minimum two auxiliary workers

are needed to support the production while supplying

the raw material and retrieving the finished products.

A manager controls the whole assembly production

according to the order and priority of product. On

each station, there is a small buffer to continue the

stream production and to overcome the bottleneck

and starving condition for the successive one.

a) Collected Data of Paper-based Assembly Process



The initial assembly process is based on paper

instruction. Different paper-based work instructions

are provided to each station. Assembly is followed by

the instructions mentioned on the paper. The papers

explain some critical steps as well. The data was

manually collected in this case, using chronometer.

The workers record the start as well as the end as

precise as possible (Table 2).

Table 2: The manual collected data

S.1 S.2 S.3 S.4 S.5 S.6

Station

Time(s/unit)

159 451 263 243 197 224

b) Collected Data of Assembly Process using IoT

Framework

After having the results of paper instructions, new

modifications are introduced on the lean assembly

line. Instructions are followed by using the IoT

Framework. All these disparate parts of the assembly

are connected via the IoT Framework including RFID

sensors. In the digitized assembly line product, flow

and layout are the same as in paper-based process.

The only parameters changed are the use of RFID

tags, display screens and paperless instructions. An

RFID system is linked with every station, which is

connected to main server and saves the progress of

the assembly. The software records the deviations and

processes the data by showing different process

parameters i.e. cycle time and lead-time. On every

station the RFID tag of a part can be read. After

successfully detecting the tag, the instructions are

displayed on a screen explaining the assembly

necessary steps. At the same time, the station records

the relevant process parameters (Table 3).

Table 3: The collected data of IOT-based process.

S.1 S.2 S.3 S.4 S.5 S.6

Station

Time(s/unit)

159 451 263 243 197 224

Takt Time: the rate at which you need to complete a

product in order to meet customer demand.

 







(1)

  3

Cycle Time: The maximum station time of an

assembly process.

  max

,

 ∈ 1…6

(2)

Cycle time for paper-based instruction is higher

than the digitalized instruction, which is mainly

caused by the work needed at station 2. Station 2

creates a bottleneck in the whole process, which leads

to high flow time and causes line balancing. Figure

11 shows the irregularity of both methods. The paper-

based method has higher irregularity than the

digitalized one. In addition, in the digitalized method,

quality checking is a pacemaker step. Pacemaker

steps create a pull flow in the whole process, which

results in less inventory and less bottlenecks.

Throughput rate: it measures the movements of

outputs within the production process.

 









(3)





Is the net available production time and  is

cycle time as explained in Equation (2).

ICINCO 2019 - 16th International Conference on Informatics in Control, Automation and Robotics

592

Figure 11: Cycle time comparison.

Eight products were assembled in one hour, but

only five products were useful, three products

defected in the case of paper-based process. On the

other hand, seventeen products were assembled and

only two of them defected. Throughput rate for

digitalized instruction is high because of less cycle

time and less bottleneck. The digitalized process has

a pacemaker with less disorder, see Figure 12.

Figure 12: Throughput rate comparison.

Flow time = total time a unit spends inside a system.









(4)

Flow time (Figure 13) also depends on the

bottleneck step. Due to bottleneck process in paper-

based method, the overall flow time is quite big.

Higher flow time results into the less production.

Figure 13: Flow time comparison.

Efficiency is very often confused with

effectiveness. Efficiency determined by the ratio of

useful output to total output. Effectiveness is being

able to achieve a desired customer demands.

 







(5)

Tw: useful total working time, Ta: net available work

time

Table 4: Efficiency comparison.

Paper-based process

IOT platform based

process

Efficiency (%) 62.63 87.5

In total, our experiment showed for that for

theconcrete application, the digitalized instructions

were more efficient than paper-based instructions, see

Figure 14.

Figure 14: Efficiency comparison.

Defected Parts: During the whole process, three

defected parts were observed in paper-based

instruction. Whereas in digitalized based method only

two parts were defected. More defected partsresult in

the unclarity of the process steps in the paper-based

instructions. Defected Products If defective articles

are not discovered in an early stage it can create

valueless work and quality shortages on products that

lead to increased cost. If the defected articles are

found before, transferring them to assembly line it

will minimize the valueless work. This can be

received by implement earlier quality controls.

5 CONCLUSIONS

Considering how industries instruct inexperienced

operators today regarding manual assembly tasks, it

is often expensive, time consuming and involve many

personnel. Especially when toady’s products become

more complex and customized, the need of operators

An IoT Framework for Assembly Tracking and Scheduling in Manufacturing SME

593

learning new manual assembly tasks increases when

having more product variants in production. If they

utilize instructions, which do not involve humans,

they often use improperly designed paper instructions

consisting mostly out of text, which is not the most

appropriate way of designing assembly instructions

considering human cognitive processes. Instructions

could be designed by using many different

technologies, but it is about how you design the

instruction that is of most importance. To fully reach

the most benefits, instructions should be effectively

designed, considering both planning and presentation

of instructions. The technology for designing

instructions should be digital, using screens or smart

tablets, to fully utilize the benefits of digitalization.

Regarding instruction performance, assembly

time and achieved product quality, of the two

instruction types (paper & digitalized), it can be

concluded that differences between the instructions

are small. The digital instruction seem to have better

performance than the other regarding both assembly

time and product quality based on our experiments.

The most impressive result of the digital instruction

was its low variation in both performance parameters,

which is reliable and consistent.

Connected to the results from the objective

quantitative measurements from the case study, with

a trend of slightly more positive results towards the

digitalized instruction regarding understanding the

instruction technology and usability during assembly.

It is therefore recommended, based on our

experiments that industries use a properly designed

digitalized instruction on a screen for inexperienced

operators, since it guides the operator how to

accurately place the hands and the technology is

familiar, easy to understand and use. Looking at the

future of manual assembly, technology within

digitalized field will be developed at a rapid pace and

will therefore be interesting to follow within the

upcoming years. Switching instruction technology

into more digital solutions is not a large investment

for companies in general, though it may have a large

impact on future business and it constitute an

opportunity for industry to reach higher

competitiveness globally and become a leader within

the digitalization field.

ACKNOWLEDGMENT

The authors would like to thank the INTERREG V A

de la Grande Région for the support of the depicted

research within the PRODPILOT project.

REFERENCES

Partha Pratim Ray. 2018. A survey on Internet of

Things architectures.

Journal of King Saud

University - Computer and Information Sciences

30, 3, 291-319.

Mohsen Darianian and Martin Peter Michael. 2008.

Smart Home MobileRFID-BasedInternet-of-Things

Systems and Services. In

2008 International Conference

on Advanced Computer Theory and Engineering

IEEE, 116–120.

Mohammad Saeid, Mohammadreza Rezvanab 2018.

Machine learning for internet of things data analysis:

a survey.

Digital Communications and Networks

4, 3

(2018), 161–175.

Jayavardhana Gubbi, Rajkumar Buyya, Slaven Marusic,

and Marimuthu Palaniswami. 2013. Internet of Things

(IoT): A vision, architectural elements, and

futuredirections.

Future Generation Computer

Systems

29, 7 (2013), 1645–1660.

Alberto Bucciero, Anna Lisa Guido 2018. Impact of RFID

and EPCglobal on Critical Processes of the Pharma

Supply Chain. Applications and Simulations, InTech,

Croatia

Xiaolin Jia, Quanyuan Feng, Taihua Fan, and Quanshui

Lei. 2012. RFID technology and its applications in

Internet of Things (IoT). In

2012 2nd International

Conferenceon Consumer Electronics,

Communications and Networks (CECNet)

. IEEE,

1282–1285.

Hermann Kopetz. 2011. Internet of Things. Springer US,

307–323.

Gerd Kortuem, Fahim Kawsar, Vasughi Sundramoorthy,

and Daniel Fitton. 2010. Smart objects as building

blocks for the Internet of things.

IEEE Internet

Computing

14, 1 (1 2010), 44–51.

Friedemann Mattern and Christian Floerkemeier. 2010.

From the Internet of Computers to the Internet of

Things. Springer Berlin Heidelberg, 242–259.

Daniele Miorandi, Sabrina Sicari, Francesco De Pellegrini,

and Imrich Chlamtac. 2012. Internet of things: Vision,

applications and research challenges. Ad Hoc Networks

10, 7 (2012), 1497–1516.

Maria R. Ebling, Roy Want 2017. Pervasive Computing

Revisited.

IEEE Pervasive Computing

, vol. 16, IEEE

Computer Society, pp 17-19.

Paul Sabanal. 2016. Thingbots: The future of botnets in

theinternet of things. RSA Conference.

Quan Z. Sheng, Xue Li, and Sherali Zeadally. 2018.

Device-Free Human Localization and Tracking with

UHF Passive RFID Tags.

Journal of Network and

Computer Applications,

Elsevier, Vol 104, pp 78-96,

Dieter Uckelmann 2016, RF-based Locating of Mobile

Objects. In: Proceedings of the 6th

International

Conference on the Internet of Things (IoT'16)

. New

York, NY, USA, ACM DL Digital Library, pp 147–

154.

Michael Kranz 2017, Building the Internet of Things,

Hoboken, New Jersey, John Wiley & Sons Inc, 2017

ICINCO 2019 - 16th International Conference on Informatics in Control, Automation and Robotics

594