Lean Human-Robot Interaction Design for the Material Supply

Process

Marco Bonini, Augusto Urru and Wolfgang Echelmeyer

ESB Logistikfabrik, Reutlingen University, Germany

Keywords: Learn HRI, QIFD, HOQI, Material Supply, Milk Run.

Abstract: Powered by e-commerce and vital in the manufacturing industry, intralogistics became an increasingly

important and labour-intensive process. In highly standardized automation-friendly environments, such as the

automotive sector, most of efficiently automatable intralogistics tasks have already been automated. Due to

aging population in EU and ergonomic regulations, the urge to automate intralogistics tasks became consistent

also where product and process standardization is lower. That is the case of the production line or cell material

supply process, where an increasing number of product variants and individually customized products

combined with the necessary ability of reacting to changes in market conditions led to smaller and more

frequent replenishment to the points of use in the production plant and to the chaotic addition of production

cells in shop floor layout. This led in turn to inevitable traffic growth with unforeseeable related delays and

increased level of safety threats and accidents. In this paper, we use the structured approach of the Quality

Interaction Function Deployment to analyse the process of supply of assembly lines, seeking the most efficient

combination of automation and manual labour, satisfying all stakeholders´ requirements. Results are presented

and discussed.

1 INTRODUCTION

In 2017, the highest robot density (that is the number

of multipurpose industrial robots in operation per

10.000 persons employed) was measured in the

automotive industry accounting for the 33% of

worldwide robot demand: in Germany, for instance,

1.162 units were installed per 10.000 automotive

employees, in the Republic of Korea 2.435 units.

When considering only general manufacturing

industry (manufacturing excluding automotive), the

numbers of units go down to 191 in Germany and 533

in Republic of Korea. The weighted average robot

density of all manufacturing industry (general plus

automotive) was assessed at 322 units in Germany

and 710 in Republic of Korea (IFR, 2018).

The high robot density in the automotive sector is

related to the high level of product and process

standardization, result of an effort accomplished in

decades in this sector, in order to reach highest

throughput and quality with minimum costs. In such

automation-friendly environments, the cost-benefit

ratio of using robotics (Bonini et al., 2015) is positive,

because of three conjoined effects: (1) increase of

throughput (capacity), (2) decrease of costs, and (3)

improvement of product quality.

With the decreasing of the standardization level of

products and processes, becomes harder to achieve a

positive impact on capacity, costs and quality (Bonini

et al., 2018). Complex tasks in less standardized

environment require robot technologies that, when

existing, are more expensive and less performing,

increasing the barrier to invest (Bonini et al., 2015).

This happens often in the general (non-automotive)

manufacturing sector, characterized by dynamic

production processes, regulated by demanding

requirements of a fast-paced global economy.

Especially the automation of the intralogistics

activities in the general manufacturing and automotive

supplier sector became challenging. Increasing number

of product variants and individually customized

products combined with the necessary ability of

reacting to changes in market conditions led to smaller

and more frequent replenishment to the points of use in

the production plant and to the chaotic addition of

production cells in shop floor layout (Urru, Bonini and

Echelmeyer, 2018). This led in turn to inevitable traffic

growth with unforeseeable related delays and increased

level of safety threats and accidents. These risks made

the logistic systems and thus the whole production

process vulnerable to inefficiency such as information

Bonini, M., Urru, A. and Echelmeyer, W.

Lean Human-Robot Interaction Design for the Material Supply Process.

DOI: 10.5220/0007966705230529

In Proceedings of the 16th International Conference on Informatics in Control, Automation and Robotics (ICINCO 2019), pages 523-529

ISBN: 978-989-758-380-3

523

loss, loss of control in the work-in-process level,

redundant inventory stored as buffer at the point of use

in the plant, missing parts, wrong parts delivered and

excessive inventories (Harris, Harris and Wilson,

2003). In logistics, the achievement of a higher robot

density has at least one additional relevance argument

and one additional criticality. The first is a

demographic component: in industrialized countries,

where quality of life is relatively high, unemployment

rate is low and population is ageing, it is becoming

increasingly hard to find labour willing to take over

ergonomically hard jobs (Abeliansky and Prettner,

2017). The second one is the impossibility of the

customer to perceive any improvement in quality due

to automation. These two argument make it at the same

time more challenging and more necessary to increase

the robot density in logistics, which is a challenge that

especially online wholesalers take really seriously.

Amazon for instance issues every year since 2015 the

“Amazon picking challenge” (Correll et al., 2018) to

stay close to the best basic-research development in

object recognition and grasping for small items of

different nature (Morrison et al., 2018). At the same

time, Amazon deployed the KIVA system on a large

scale in its distribution centres and warehouses. This

automates the transport functionality of the

commissioning process using high performance

available technology, while leaving the unstructured

task of the picking to a human operator (Li, 2016).

Recent research (Bonini and Echelmeyer, 2018;

Bonini, Urru and Echelmeyer, 2019) focuses on

formalizing this empirical process of finding the right

level of automation. Answering in a structured way to

the question “who-does-what” between man and

automation could be the key leading to lean human-

robot interaction, thus increasing the robot density

even in the logistic sector, with a substantial relief for

human operators of ergonomically hard tasks. Using

the structured approach provided by Bonini et al.

(Bonini, Urru and Echelmeyer, 2019), in this paper we

analyse the process of supply of assembly lines,

seeking the most efficient combination of automation

and manual labour, satisfying all stakeholders´

requirements. After a brief summary of the state of the

art for allocation of functionalities between human and

automation, with a specific focus on the Quality

Interaction Function Deployment (QIFD

) method for

lean HRI, we present the scenario and the result of the

application the QIFD, which are then discussed.

2 STATE OF THE ART

As fully autonomous systems are often too expensive

and low performing and simpler cheaper systems are

not enough flexible, Bonini et al. (Bonini, Urru and

Echelmeyer, 2019) proposed to set the focus on using

simpler cheaper systems in interaction with human

operators. If the interaction is well designed, this

could improve costs, performances and acceptance.

In order to find convenient balance between manual

work and automation solutions, first the so-called

“all-or-non-fallacy”, namely the false idea that either

a process should be fully automated, or it should be

fully manual (Sheridan and Verplank, 1978), needs to

be abandoned. This presumes an allocation of

functions among automated and human agents that

can follow several principles, the simplest of which is

the Fitts´ list “Men are better at-Machines are better

at” (MABA-MABA) (Fitts, 1951) updated through

the years as new technologies were released (Price,

1985; Hancock and Scallen, 1998). More elaborated

qualitative and quantitative approaches are those of

the comparative, leftovers and economic allocation

(Rouse, 1991) or the sharing of control (Inagaki,

2003). Most of these methods approach heuristically

the function allocation problem, delivering results

that need to be validated. Others (Ranz, Hummel and

Sihn, 2017) developed analytic approaches aimed to

objectivize the function allocation problem by

seeking an optimal solution. While effective for a

narrow and specific low-level task of the work

breakdown structure, these kind of analytic optimum-

seeking approaches are ill suited for the analysis of a

large process chain, where too many dynamic

parameters come at play. The problem with existing

methods from the literature is that they are either

exclusively qualitative, or, in the effort to quantify the

decision making process, focus on a narrow array of

parameters. For this reason, with the objective of

function allocation in the line supply process, in this

paper we use the alternative approach introduced in

(Bonini and Echelmeyer, 2018) and refined in

(Bonini, Urru and Echelmeyer, 2019), namely a 12-

steps heuristic method that functions as a decisional

support for process design. The method has been

applied in a focus group, where participants had

various competences. The decisional process has

been tracked and documented using the House of

Quality Interaction visual tool. Thanks to the QIFD

method, different automation scenarios were created

and evaluated with respect to their compliancy to two

sets of requirements of all process stakeholders: (1)

hard requirements, representing the view of the

investors and considering parameters such as the need

for automation, efficiency and performance and (2)

soft requirements, representing the view of the

user/partner of the automation, thus considering

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

524

parameters such as ergonomics, complexity, work

balance, accountability and acceptance. The 12 steps

of the method are in order: (1) eliciting and weighting

of hard requirements, (2) identification of needed

functionalities, (3) evaluation of impact of

functionalities on compliancy to requirements, (4)

identification of synergies and conflicts in automation

of functionalities, (5) calculation of utility of

functionalities, (6) estimation of relative complexity

of automation and relative complexity of manual

execution of each functionality, (7) calculation of

convenience of automation for each functionality, (8)

creation of automation scenarios based on

convenience and synergies or conflicts, (9) estimation

of compliancy of each scenario to hard requirements,

(10) estimation of relation between functionalities

and the perception-action model (Parasuraman,

Sheridan and Wickens, 2000), (11) calculation of

compliancy of each scenario to soft requirements,

such as complexity of automation, ergonomics,

workload balance (mental and physical),

accountability and acceptance, (12) calculation of

total requirement compliancy score from compliancy

to soft and hardware requirements (Bonini, Urru and

Echelmeyer, 2019). The application of this method

(section 4) to the scenario described in section 3 lead

to novel results, namely to an innovative human-robot

interaction approach for the material supply process,

presented in section 5 and discussed in section 6.

3 SCENARIO DESCRIPTION

Being the scope of the investigation the material

supply process in a generic production plant, in this

section, a typical scenario is introduced. As illustrated

in Figure 1, the transportation of material in

production logistics takes place between 4 main

areas: (1) warehouse, (2) supermarket (SM) work cell

or station (WS) and (4) outbound. After the inbound

process different kinds of unit load (UL), such as for

example coils, boxes, mesh cages and pallets, are

stored in the warehouse. From the warehouse, the

goods are then transported to the supermarket where

they are prepared to be delivered to the work cell

Usually, in this phase, the goods are bundled in

standardized unit loads according to the company best

practice. The selectivity in the supermarket is often

high, meaning that the unit load, which contains

homogenous goods, could be individually handled.

According to the needs, in unit load, of each work

cell, a transportation order list is compiled. The unit

loads are commissioned in the super market and

transported to the related work cell. Once the material

has been utilized, the empty unit load (i.e. euro

container) needs to be transported back to the SM, so

that the cycle could start again. For the sake of

completeness, in Figure 1 also transportation of work

in progress (WIP) between work cells and of the final

products to the outbound area are represented.

The analysis focuses both information and

material flow between supermarket and work cell.

Moreover, to simplify the method implementation,

we will consider a scenario where the supermarket is

dedicated to only one unit load: euro container. The

full euro containers are commissioned in the

supermarket and transported to the workstation by

means of manual transport/push carts or forklifts. The

transport cart is pushed to the nearest reachable area

(H), where the euro container is actually needed, the

point of use (PoU), as shown in Figure 2. Once the

area H is reached, the operator manually picks the full

euro container and transports it to the PoU. As the

full euro container is positioned in the shelf, the

empty one could be brought back to the push cart.

This process is repeated until all the euro containers

on the push cart are delivered. The empty euro

containers collected during the delivery tour are then

brought back to the SM and the commissioning

process into the SM starts again.

On the basis of this scenario, in the next section,

the design of a lean Human-Robot Interaction for the

material supply process will be presented.

4 METHOD APPLICATION

In this section, for each step of the Quality Interaction

Function Deployment presented in the state of the art,

a brief description of the main results will be given.

Figure 1: Example of material flow in production logistics.

Lean Human-Robot Interaction Design for the Material Supply Process

525

Figure 2: Material flow between H and PoU.

For the sake of conciseness, some steps of the QIFD

method are here reported as grouped (e.g. 4.3 Impact,

Correlation and Utility).

4.1 Hard Requirements

In this first step of the heuristic method, hard process

requirements are investigated. Given the dynamicity

of a modern production process, as already mentioned

in the introduction, and the general goal improving

the processes through automation, a list of 13

requirements has been identified. The requirements

have been then analyzed and ordered by importance.

To avoid influencing the final result with personal

opinions and believes on automation priorities, the

same weight has been assigned to all the requirements

concerning automation.

The resulting requirements ranking with the

related normalized importance (in brackets) follows:

(1) the system must efficiently answer throughput

changes (0,11), (2) the system must be able to

efficiently answer to changes in the layout (0,10), (3)

the system must generate a low traffic (0,10), (4) the

system must be scalable (0,09), (5) the system must

be able to handle different kind of standard UL (0,08),

(6) the ordering of full UL (0,06), (7) the

commissioning of full UL in the SM (0,06), (8) the

preloading of full UL (0,06), (9) the loading of full

UL (0,06), (10) the transport of UL between

Supermarket and H (0,06) (11) the transport of UL

between H and PoU (0,06) (12) the exchange full-

empty UL (0,06) and (13) the unloading of empty UL

(0,06), must be automated.

4.2 Functionalities

After eliciting the requirements, functionalities are to

be deployed and divided to the atomic level in which

they could be assigned to either the human or the

automation. The list of identified functionalities is

hereafter given: (1) ordering of full UL, (2)

commissioning of full UL, (3) preloading of full UL,

(4) loading of full UL, (5) transport of UL between

SM and different H, (6) navigation between SM and

different H, (7) transport of UL between H and PoU,

(8) navigation between H and PoU, (9) exchange

full/empty UL at the PoU, (10) Unloading of empty

UL.

4.3 Impact, Correlation and Utility

Once the impact of hard requirements on each

functionality has been assessed by means of a

logarithmic scale and the correlation between

automation of different functionalities have been

evaluated, the relative utility of each functionality

have been calculated. The functionalities with the

highest utility are: 1) ordering of full UL, 2)

commissioning of full UL, 3) preloading of full UL.

While the functionality with the lower utility is the

transport of UL between SM and different H.

4.4 Complexity

In this step the complexity is evaluated from two

different point of view: 1) the complexity of

automation and 2) the relative complexity the human

operator encounters in the manual execution of a

functionality. For instance the most complex

functionality to be automated turns out to be the

exchange full/empty UL at the PoU, while the most

complex and strenuous functionality, currently

manually executed by the operator, appears to be the

commissioning of full UL.

4.5 Convenience

Estimating the convenience considers both the

potential benefits of automating high impact

functionalities, even if their automation has a high

complexity level, and the possibility of automating

low impact functionalities, when their automation is

extremely simple (low complexity). The functiona-

lities are ranked according to their convenience. This

ranking will be input for the next step of scenarios

development. The most convenient functionalities to

be automated are: 1) ordering of full UL, 2)

commissioning of full UL and 3) the transport of UL

between SM and different H.

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

526

4.6 Scenarios

Overall 12 different scenarios are identified. Starting

from a fully manual scenario, functionalities have

been assigned to automation according to the

convenience and the correlation matrix. For each

scenario automated and manual processes are

described together with the needed technologies.

4.7 Compliancy Hard and Soft

Requirements

In the following method step scenarios are evaluated

against hard and soft requirements. The hard

requirements are the ones identified in the beginning,

while the soft ones encompass (1) complexity of

scenario automation, (2) ergonomics, (3) mental work

balance, (4) physical work balance, (5) accountability

and (6) acceptance. To each scenario a weighted score

is assigned. Scenarios are ordered in a ranking

according to the score. In the following section, the

first three scenarios of the ranking will be presented.

5 RESULTS

The top tier scenarios will be hereafter described

considering three main aspects: (1) functionalities

allocation, (2) technologies and (3) processes.

In Table 2 functionalities are assigned according

to the scenario to the worker (M - manual) or to the

automated solution (A - automation). As noticeable in

Table 2, some of the functionalities are either manual

or automated, independent of the scenario (greyed out

rows).

For each functionality to be automated, the

appropriate technology should be chosen. A summary

of the technologies chosen is introduced in the

following Table 1. The logistic system concept for the

material supply is based on tugger trains. Thanks to

the information system and sensors available at the

POU, the material is directly ordered at the POU and

the picking list automatically compiled. The picking

Table 1: Functionalities and technologies.

Table 2: Scenarios and functionalities allocation.

list is fed to an automated storage and retrieval system

(AS/RS), which will take over the commissioning

process of the needed ULs. The UL will be then

automatically sorted and prepared for the loading on

the tugger train by means of a Driving-Thru loading

concept, developed by the Technische Universität

München (Dewitz, Galka and Günthner, 2012). The

automatic loading is only foreseen for the scenario1.

In scenario 2 and 3 the loading of the full UL is done

manually. The functionality of transport of UL

between stations is always accomplished by means of

a tugger train, driven by an operator. Once the tugger

train has reached the target station H, the UL should

be transported to the POU as shown in Figure 2. An

additional trailer should be considered, in order to

transport a mobile platform, which is able to carry

more than one UL at a time and to follow the operator

up to the POU. At the POU the operator will exchange

the full UL with the empty one, then the mobile

platform (AGV) will follow the operator back to the

tugger train. After delivering all the UL, the operator

drives back to the SM where in the case of automated

functionality (scenario 1 and 3) the unloading of

empty UL could be accomplished by the same

technology Driving-Thru loading concept. The tugger

train is ready to start a new delivery cycle. In Figure

3 the scenario 1 is graphically described. Comparing

this scenario with the initial fully manual one, it is

noticeable how the number of transportation needed

Figure 3: Lean Human-Robot Interaction concept for the

material supply process.

1 Ordering of full UL A

2 Commissioning of full UL A

3 Preloading of full UL A

4 Loading of full UL A

5 Trasnport of UL between stations A

7 Trasnport of UL between station and POU A

10 Unloading of empty UL A

Drive-Thru Loading Concept

Tugger Train

Mobile platform (AGV)

Drive-Thru Loading Concept

Functionality

Direct order, information system and

sensors at the PO

Aautomated Storage and Retrieval

System or simila

Technology

Functionality

Scenario

1Ordering of full UL

AAA

2 Commissioning of full UL

AAA

3 Preloading of full UL

AAA

4 Loading of full UL

AMM

5 Trasnport of UL between stations

AAA

6 Navigation between stations

MMM

7 Trasnport of UL between station and POU

AMA

8 Navigation between station and POU

MMM

9 Exchange full/empty UL

MMM

10 Unloading of empty UL

AMA

Lean Human-Robot Interaction Design for the Material Supply Process

527

to fulfil the overall material need is drastically

decreased. Mainly thanks to the advantages offered

by the two transportation technology chosen: the

tugger train and the mobile platform (AGV). the

introduction of an AGV able to carry more than one

full UL at a time reduces the empty travel, increasing

the overall system efficiency.

6 DISCUSSION

The top tier scenario of the 12 developed logistic

concepts are in this section briefly discussed.

The most promising scenario, scenario 1, is the

best in ergonomics due to the automation of loading

of full UL, of the transportation between station H

and PoU and of the unloading of empty UL. The

acceptance of the scenario ranks also as the best,

thanks to the active involvement of the operator.

Technologies available for the implementation of this

scenario will be ripe enough to be integrated in a

whole system with a short-term horizon (within about

three years). Development effort to adapt the AGV

platform is estimated to be low.

The second ranked scenario, scenario 2, could be

implemented without any development effort, since

the loading of full UL, the unloading of empty UL and

the transportation of UL between station H and PoU

is manual. Within this scenario, only the most

impacting processes are object of automation, i.e. the

processes of ordering, commissioning, pre-loading

process and that of physical transportation from the

SM to the stops of the route H (but not the process of

navigation).

The third ranked scenario, scenario 3, differs from

the previous one only in two aspects. The transport of

UL between H and PoU, manual in scenario 2, is now

automated as is the unloading of empty UL, manual

in scenario 2. With respect to scenario 1, the loading

of full UL is here manual instead of automated; this

creates a disadvantage concerning ergonomics, but an

advantage concerning the smaller impact on exiting

layout making it overall a less investment-intensive

logistic concept.

It is important to remind that these results were

achieved using a heuristic method and should not be

considered as optimal, but rather as the best

achievable result of the competences and discussion

of the participants to the focus group. This means that

different participant with different background and

knowledge could have, for instance, chosen different

technologies. Moreover, different focus groups and

different application contexts could lead to a different

interpretation of hard and/or soft requirements, with a

non-negligible impact on their ranking/weighting

process. This could lead to substantially different

logistic concepts, compared with the ones presented

in this paper. For these reasons, in the forthcoming

research, results should be validated with an

economic convenience analysis. The analysis should

aim at estimating and assessing the economic effort

to implement each different logistic concept,

providing an additional criteria for the overall concept

evaluation.

7 CONCLUSION

After the explanation of the relevance of the topic and

an overview on the state of the art in function

allocations among automated and human agents, in

this paper we used the structured approach of the

Quality Interaction Function Deployment to analyse

the process of material supply in production

environments. Applying this method, we designed 12

automation scenarios that were evaluated and

discussed with respect to their compliancy to two sets

of requirements: (1) hard requirements, representing

the view of the investors and considering parameters

such as the need for automation, efficiency and

performance and (2) soft requirements, representing

the view of the user/partner of the automation, thus

considering parameters such as ergonomics,

complexity, work balance, accountability and

acceptance. The three top tier of the 12 scenarios were

presented and discussed. Considering the currently

available technologies, the most promising logistic

concept for the automation of the material supply

process in production environments envisions a lean

human-robot interaction with the automation of all

activities, except the navigation and exchange of

full/empty unit loads at the point of use, which are

still being assigned to the human operator.

ACKNOWLEDGEMENTS

Supported by „EFRE Program Baden-Württemberg

2014-2020“, Project: ZAFH Intralogistik.

REFERENCES

Abeliansky, A. and Prettner, K. (2017) ‘Automation and

Demographic Change’, Ssrn, (310). doi:

10.2139/ssrn.2959977.

Bonini, M. et al. (2015) ‘Towards the full automation of

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

528

distribution centers’, in 2015 4th IEEE International

Conference on Advanced Logistics and Transport,

IEEE ICALT 2015. doi: 10.1109/ICAdLT.2015.7136

589.

Bonini, M. et al. (2018) ‘Automation of Intralogistic

Processes through Flexibilisation - A Method for the

Flexible Configuration and Evaluation of Systems of

Systems’, in ICINCO 2018 - Proceedings of the 15th

International Conference on Informatics in Control,

Automation and Robotics, pp. 390–398. doi:

10.5220/0006878003900398.

Bonini, M. and Echelmeyer, W. (2018) ‘A method for the

design of lean human-robot interaction’, Proceedings -

2018 11th International Conference on Human System

Interaction, HSI 2018, pp. 457–464. doi: 10.1109/

HSI.2018.8430879.

Bonini, M., Urru, A. and Echelmeyer, W. (2019) ‘The

Quality Interaction Function Deployment for lean

Human-Robot Interaction’, in Accepted for

presentation at the 24th International Conference on

Methods and Models in Automation and Robotics.

Correll, N. et al. (2018) ‘Analysis and Observations From

the First Amazon Picking Challenge’, IEEE

Transactions on Automation Science and Engineering,

15(1), pp. 172–188. doi: 10.1109/TASE.2016.2600527.

Dewitz, M., Galka, S. and Günthner, W. A. (2012) ‘Drive-

Thru Loading Concept for In – Plant Milk Runs’, in

Proceedings of XX International Conference MHCL

´12, pp. 6–6.

Fitts, P. M. (1951) ‘Human engineering for an effective air-

navigation and traffic-control system.’ National

Research Council, Div. of.

Hancock, P. A. and Scallen, S. F. (1998) ‘Allocating

functions in human--machine systems.’ American

Psychological Association.

Harris, R., Harris, C. and Wilson, E. (2003) Making

Materials Flow. Cambridge, MA, USA: The lean

Enterprise Institute.

IFR (2018) Executive Summary World Robotics 2018

Industrial Robots.

Inagaki, T. (2003) ‘Adaptive Automation: Sharing and

Trading of Control’, Handbook of Cognitive Task

Design, pp. 147–169. doi: 10.1201/9781410607775.

Li, J. (2016) ‘Design Optimization of Amazon Robotics’,

Automation, Control and Intelligent Systems, 4(2), p.

48. doi: 10.11648/j.acis.20160402.17.

Morrison, D. et al. (2018) ‘Cartman: The Low-Cost

Cartesian Manipulator that Won the Amazon Robotics

Challenge’, in 2018 IEEE International Conference on

Robotics and Automation (ICRA), pp. 7757–7764. doi:

10.1109/ICRA.2018.8463191.

Parasuraman, R., Sheridan, T. B. and Wickens, C. D. (2000)

‘A model for types and levels of human interaction with

automation’, IEEE Transactions on Systems, Man, and

Cybernetics Part A:Systems and Humans., 30(3), pp.

286–297. doi: 10.1109/3468.844354.

Price, H. E. (1985) ‘The allocation of functions in systems’,

Human factors. SAGE Publications Sage CA: Los

Angeles, CA, 27(1), pp. 33–45.

Ranz, F., Hummel, V. and Sihn, W. (2017) ‘Capability-

based Task Allocation in Human-robot Collaboration’,

Procedia Manufacturing

. The Author(s), 9, pp. 182–

189. doi: 10.1016/j.promfg.2017.04.011.

Rouse, W. B. (1991) ‘Design for success: a human-centered

approach to designing successful products and

systems’. New York: Wiley (Wiley series in systems

engineering), pp. xv, 287 p.

Sheridan, T. B. and Verplank, W. L. (1978) Human and

Computer Control of Undersea Teleoperators, MIT.

Cambridge: MIT.

Urru, A., Bonini, M. and Echelmeyer, W. (2018) ‘Planning

and dimensioning of a milk-run transportation system

considering the actual line consumption’, IFAC-

PapersOnLine, 51(9), pp. 404–409. doi: 10.1016/

j.ifacol.2018.07.066.

Lean Human-Robot Interaction Design for the Material Supply Process

529