Method for Design of Human-industrial Robot ... - Science Direct

Available online at www.sciencedirect.com

ScienceDirect Procedia Manufacturing 11 (2017) 4 – 12

27th International Conference on Flexible Automation and Intelligent Manufacturing, FAIM2017, 27-30 June 2017, Modena, Italy

Method for design of human-industrial robot collaboration workstations Fredrik Orea,b*, Lars Hanssonb,c,d, Magnus Wiktorssona a

Mälardalen University, School of Innovation, Design and Engineering, Eskilstuna, Sweden b Scania CV AB, Global Industrial Development, Södertälje, Sweden c University of Skövde, School of Engineering Science, Skövde, Sweden d Chalmers University of Technology, Department of Product and Production Development, Gothenburg, Sweden

Abstract In order to fully utilise a 3D simulation software capable of evaluating hand-guided human-industrial robot collaborative (HIRC) work tasks, there is a need of a HIRC design process for early production development stages. This paper proposes a HIRC design method that uses the possibilities of the demonstrator software in the HIRC workstation design process. The method is based on Pahl and Beitz’s engineering design method; it interprets all their phases and activities into HIRC design-specific ones. © 2017 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license © 2017 The Authors. Published by Elsevier B.V. (http://creativecommons.org/licenses/by-nc-nd/4.0/). Peer-review under responsibility of the scientific committee of International the 27th International Flexible Automation and Peer-review under responsibility of the scientific committee of the 27th ConferenceConference on Flexibleon Automation and IntelligentManufacturing Manufacturing. Intelligent Keywords: Human-robot collaboration; Human-robot Interaction; Simulation; Collaborative assembly; Hybrid workstation; Workstation design; Digital human modelling; Design method

1. Introduction Human-industrial robot collaboration (HIRC) is a rapidly growing field in research. A literature search performed in 2015 showed a four-time increase in the number of academic publications per year between 2005 and 2014 [1]. The potential to combine the beneficial characteristics of the human with those of the industrial robot opens huge possibilities to simultaneously increase productivity and reduce ergonomically bad work postures [2]. The vision is * Corresponding author. Tel.: +46-8-553-51237. E-mail address: Fredrik.ore@scania.com

2351-9789 © 2017 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). Peer-review under responsibility of the scientific committee of the 27th International Conference on Flexible Automation and Intelligent Manufacturing doi:10.1016/j.promfg.2017.07.112

Fredrik Ore et al. / Procedia Manufacturing 11 (2017) 4 – 12

to create more productive workstations through combining human intelligence and flexibility with industrial robotic strength, endurance and accuracy [3, 4]. Simulation software is used in manufacturing companies early in production development processes to shorten development time, increase quality and reduce costs [5]. These tools are used to support decision making in the companies [6] and are an integral part of the engineering activities in many manufacturing companies [7]. However, this statement is not valid for HIRC systems; the possibility to virtually evaluate entire HIRC systems before making an investment decision is very limited in available simulation software. In order to meet this need, a demonstrator simulation software was developed, making it possible to design and evaluate HIRC workstation layouts early in the production development phases [8]. This software enables simulation of hand-guiding HIRC tasks in 3D CAD environments. It can be used to analyse reachability for both industrial robots and humans, present layout alternatives and be a tool for risk assessment in HIRC workstation design assignments. The software generates quantitative outputs considering operation time and biomechanical load assessments for fully manual, fully automated and any hybrid (HIRC) workstation. These quantitative outputs can be used to compare alternative solutions in an objective way. The demonstrator software is designed to be used during HIRC workstation design. However, to get the full potential from virtual simulations, there is a need of a HIRC design process for early production development stages. Thus, the aim of this paper is to propose a HIRC design method that can be used in early phases on production development processes. 2. Method Experience from design of four industrial HIRC cases [8, 9] in the development of the HIRC demonstrator simulation software led to the development of the proposed method. In order to gain validity of the proposed HIRC design method, it was connected to existing design methods. Pahl and Beitz’s engineering design method was chosen, as it has become the reference work to teach design engineers a systematic method to include a heterogeneous set of theories and methods for one product design process [10]. The latest English edition (3rd) of their work (co-authored by Feldheusen and Grote) was used in developing the HIRC design method proposed in this paper [11]. The method is referred to as the engineering design framework in this paper. 3. Frame of reference Systematic design has been developed over the past several decades as a best practice for product design [12]. A frequent view in design research is to consider the design process as following a sequential scheme. From product planning and design, a number of processes have been proposed that follow that scheme, e.g., [11], [13-15]. These generic design processes have been applied in production development processes, such as [16-19]. The book Engineering design: a systematic approach written by Pahl and Beitz was first released in 1977 in German, Konstruktionslehre [20]; it has become the reference work in teaching design processes [10]. This engineering design framework consists of four main phases: planning, conceptual design, embodiment design and detailed design [11]. Despite the linear flow from planning to detailed design, it is important to understand that design is an iterative process, demanding use of new knowledge back in previous phases and activities. HIRC design methods presented in research publications are mainly limited to the work task allocation problem, i.e., which resource is most suitable to perform a task: the human or the industrial robot? Pini el al. also base their design approach in the engineering design framework presented by Pahl and Beitz [21]. They present qualitative suitability comparison between manual and robotic task allocation. Economic profitability of the solution are included. Chen et al. present a method to use multi-objective optimisation techniques to choose a suitable task allocation based on assembly time and economic cost [22]. Tsarouchi et al. have a similar approach in their task allocation method [23]. They present an decision making method that considers mean flowtime and utilisation costs in its decisions. All these methods use time and cost as evaluation criteria, but none of them describes how to gather data into the selection process. One approach is to measure these before the task allocation can begin. However, in early phases of production design it is difficult to achieve these data since no physical workstation exists. This

5

6


highlights the need of simulation software in order to make accurate production investment decisions early in the production development process. A few early developments of HIRC simulation software have been presented. The challenging part in these tools is to simulate human motions based on digital human models, so-called manikins. The majority of the publications presented include a lightweight model of the human [24, 25], often a skeleton model of a human that is fed with motion capture data in order to move in a representative manner. These manikins interact with industrial robots in a 3D CAD environment. Other authors include more advanced digital human modelling software in the simulations in order to predict human motions without the need of motion capture data. Tsarouchi et al. use the off-line programming software “Process Simulate” to simulate the human and industrial robotic tasks [26]. To the knowledge of the authors of this paper, none of the existing digital human modelling software has been able to simulate simultaneous human robot motions on an object in a HIRC workstation. This feature is, however, included in the HIRC software presented by Ore et al. [8], where the digital human modelling software IMMA is used to represent human motions. This software is used in the proposed HIRC design method, and is referred to as the HIRC simulation software in this paper. 4. Proposed method for HIRC workstation design The proposed HIRC design method based on the engineering design framework is presented in Fig. 1 with the four phases and their corresponding activities. These phases and activities are described below. The input to start the HIRC design method is an identified workstation, existing or in the planning phases, that might benefit from an industrial robot in performing some work tasks. The identified workstation must have geometric boundaries on the shop floor as well as defined functional limitations in terms of what to produce. 4.1. Planning and clarifying the task The term “task” is in this headline used in the meaning of the general assignment for the HIRC workstation. The goal of this phase is to gather needed information about the assignment, to be used in the following phases. The outcome is a requirements list that specifies the workstation design problem, including its constraints and objectives. The first activity, HIRC possibilities of investigated workstation, includes a rough analysis of the HIRC potential in the workstation. This is based on fundamental knowledge of the workstation and characteristics of industrial robots. The workstation is described by its functional characteristics (what to produce), geometrical characteristics (positions and constraints of all objects) and expected output (operation times and volumes). The second activity is analyse company-specific demands on workstation. There might be constraints limiting what industrial robot to select. This kind of analysis also includes identifying who will work in the stations. The anthropometric measurements differ between humans, and there are different databases with different nationality, gender, etc., of the operators, and this has to be considered in order to design ergonomically friendly workstations. Company-specific safety demands have also to be identified and considered here. The next activity is formulate evaluation criteria and design variables. The HIRC simulation software evaluates the workstation based on operation time and biomechanical load. Many variables in the workstation design influence these criteria, e.g., the industrial robot variant and position in the layout, the design of the industrial robot gripper as well as positions of all materials in the station. Other workstation characteristics (e.g., hand-over positions between industrial robot and humans, positions of buttons and other operator interfaces) are also of importance. On top of this is also the selected anthropometric database used to get appropriate measurements of the humans in the simulation. In total, six design variables are proposed in the future workstation design: industrial robot variant, industrial robot position, industrial robot gripper design, material position, workstation equipment position and anthropometric database. The first evaluation criterion “operation time” includes both industrial robotic and manual times. The time of the industrial robot tasks is calculated using industrial robot and environmental position data. Human operation time is calculated using a predetermined motion time system [27]. The second evaluation criterion, “biomechanical load”, is measured using the evaluation method of rapid upper limb assessment (RULA) [28]. RULA investigates the musculoskeletal injury risk for humans by evaluating the individual poses and assessing the injury risks of those positions on the human body on a risk level from one to seven, where a high score indicates

7


a high risk of future injuries. [8] is recommended for a more thorough explanation of how these two evaluation criteria are captured in the demonstrator software. """% !"" !!"!$!""% !"" &!&'!!% !"" #"$#" " !$ ! "$#" " !$ ! " # "!!"

"&!!" &!" %#" $#""$ "!!"" "

!#"

&!% !"" $% !""! !$ !%!"!

" &% !""#""

$

# "!!"

"$!#" "$#"#!""!" !#"#"" !""#"" !""! Fig. 1. Proposed HIRC design method inspired by Pahl and Beitz’s engineering design method [11].

When the evaluation criteria and design variables of the HIRC design problem have been established, it is time to set evaluation criteria and design constraints. The objective is defined in terms of the evaluation criteria, and any constraints on the objective have to be defined, e.g., maximum limit on available operating time or allowed biomechanical load score. Two dimensions in the objective generate a multi-objective problem, thus the relative weight of operation time and biomechanical load has to be decided. Numerical design constraints on the six variables are set to be used in the design phase. In this early stage of the HIRC design method it is important not to be too detailed in setting the design constraints; these can be limited further in future phases and activities. The last activity in the planning phase is elaborate a requirements list. This list includes results from the activities in this phase presented as design specifications to be used in the following workstation design.

8


4.2. Conceptual design The conceptual design phase aims to produce the most appropriate principal solution to the workstation design problem. This requires creativity of the designer in order to search beyond traditional methods and techniques to complete the assignment and to consider new ideas and problem descriptions to solve the problem in an optimal way. The first activity identify essential problem is a crucial part of identifying innovative solutions. It includes an abstraction of the workstation design problem “to find the crux of the task” [11, p. 161]. The goal is to broaden the problem formulation to find a technology independent description of the essential function. The requirements list developed from previous phases is abstracted by eliminating personal preferences, focusing on the essential function, reducing qualitative data to essential statements and later formulating them in solution-neutral ways [11, p. 165]. The next activity, model and analyse material flow in functional block diagram, includes creating a block diagram showing functions and sub-functions of the problem. Creativity from the designers (preferably a group of individuals) is then used to find new solutions in order to fulfil the functions. These solutions can include new techniques as well as changes in the order of executing the sub-functions in order to get optimal workstations. The final activity, evaluate technical variants against technical and economic criteria, includes summarising the ideas to practical concepts and evaluate them. The final concepts are evaluated based on the criteria defined in the planning phase. Even though it might be difficult to get accurate data on the concepts as they are often vague, they are evaluated relative to each other to find the best principal solution to the workstation design problem. 4.3. Embodiment design The third phase is the embodiment design phase. It includes the actual HIRC workstation design. The first activity, preliminary design of HIRC workstation, includes the first design cycle, where the variables from the planning phase are developed to discrete values and the HIRC workstation design problem is evaluated in order to produce preliminary solutions. The values are set within the design constraints with a coarse grid that is detailed in later activities. The industrial robot variant and anthropometric database are selected based on the company-specific demands on the workstation. The other four variables are workstation-specific and values have to be decided. Some of these six variables might be set as constants in the case investigated. The first part of this activity is to limit the potential solutions to feasible ones. In the HIRC simulation software it is possible to simulate any kind of workstation. In practice, however, there are many human and automation constraints that limit the space of possible HIRC solutions. The HIRC system should combine what the human lacks in strength, endurance and accuracy with industrial robots and include human intelligence and flexibility that the industrial robot traditionally lacks. In previous work, Ore et al. identified automation constraints for a specific manufacturing industry [29]. The paper presented a process to create a task allocation between industrial robot and human considering automation constraints. This process can also be used with manual constraints in order to define a task allocation between industrial robot and the human; manual constraints are also included in the proposed HIRC design method. It includes a hierarchal task analysis (HTA) of the workstation that breaks the overall task down to manageable tasks in order to find goals and sub-goals of a system [30]. These tasks are then examined considering automation and human constraints in order to identify what resources are capable of performing a task; they are presented in the form of a resource allocation table, exemplified in Table 1. This resource allocation is later used to select the most appropriate task allocation [31]. Table 1. Example of resource allocation. These alternatives give 4 possible solutions (2 x 2). Work task 1. Get object 2. Position object in fixture 3. Tighten nuts 4. Move final product to pallet

Ind. robot X X

Human X X X X

9


The variables from the planning phase must also be considered. These are combined into a morphological matrix that presents the possible discrete values that these variables can take, which are later combined into unique solutions, as exemplified in Table 2. It is important to ensure that only technically feasible solutions are considered in this stage as the higher number of alternatives quickly increases the solution space and time for the next step, the simulation. Tables 1 and 2 give the total number of simulations to perform through the HIRC simulation software (in these examples 4 x 128 = 512). These simulations are performed automatically through an interface where the design variables and task allocations are altered in the software. This step generates operation time and RULA values each of these simulations and enables evaluation of what task allocation and design variables are the most suitable for the workstation. One or many primary solutions are chosen to be further evaluated. Table 2. Example of morphological matrix. These alternatives give 128 possible solutions (4 x 8 x 4). Variables Industrial robot variant Industrial robot position Industrial robot gripper design Material position(s) Workstation equipment position(s) Anthropometric database

Alternatives R1 R2 A1 A2 RG1 MP1 WE1 WE2 M1

R3 A3

R4 A4

WE3

WE4

A5

A6

A7

A8

Refine and improve HIRC workstation design is the second activity in the HIRC design method. It includes the next design loop and further elaboration through simulation of the primary solutions selected. The variables are here further evaluated through a finer grid showing the geometric positions of industrial robot, material and other workstation equipment, as well as a more detailed selection of the industrial robot variants and grippers to be used. If needed, also the digital manikins used in the simulation (anthropometric database) are presented in a more detailed and appropriate manner. Here the task allocation from previous evaluations should also be reconsidered, if needed. This design loop generates a new optimal preliminary HIRC workstation design that is chosen for the next steps. The next activity, find and solve errors and weak spots, includes activities performed to ensure that the selected preliminary design is evaluated properly and that no errors and weak spots are carried on to the next phase. There is no clear method presented for this work, but a critical evaluation of the results from the simulation is required. The last activity, prepare the preliminary HIRC workstation documentation, includes summarising the design into CAD models or sketches and presenting the quantitative outputs from the HIRC simulation software in a suitable way. In addition to the workstation layout, other results are also generated from the software, such as industrial robotic program code and human work instructions. The industrial robot code is saved in a neutral language enabling import to the physical industrial robot. The human work instructions, based on the task allocation and the human motions visible in the software, are used to create instructions in any standard desired by the manufacturing company. 4.4. Detail design The activities in the last phase of the HIRC design method includes detail evaluation and adjustments of the resulting documentation from the work. The workstation layout is saved in a form that suits the company in question, in a 3D CAD model in a neutral format or in 2D printed drawings. The industrial robot program code and the human work instructions are developed to be used in practice. The industrial robot code is validated by offline evaluation in industrial robot-specific software to ensure that the industrial robot motions correspond with those from the HIRC demonstrator software. The human work instructions are validated through company internal work instruction standards and through reviews from experienced operators.

10


5. Discussion 5.1. Clarifying terminology of the engineering design method Some terms in the engineering design framework need to be clarified in the HIRC workstation design context. First, the term task in the first phase “planning and clarifying the task”. Task shall here be used in the meaning of assignment and not as in a work task allocation problem (mentioned in the frame of reference chapter). Another term is layout. This term is used in the engineering design framework with a slightly different meaning from the general one in workstation design. Pahl et al. define “layout” as the general arrangement of objects, while “form design” contains the accrual shape and material of the object [11, p. 237]. In the early phases of the engineering design framework the focus is on layout, and as time passes, the focus gradually turns towards form design. The physical layout in workstation design should not be confused with the layout term in the engineering design framework. 5.2. HIRC design method The HIRC design method is presented as a linear process, both in Fig. 1 and in this paper. This gives the impression that the process is an easy-to-follow, step-by-step method in order to reach an optimal solution. However, it is important to consider the upgrade and improve activity presented as feedback loops to the right in in the method in Fig. 1. These represent the possibility and need to reconsider decisions from previous phases and activities. Learnings from later phases in the process might be fed back into earlier activities. Specifically, two design loops are presented in the embodiment phase; however these could very well be three or four as well as one, all depending on the workstation design problem. In some HIRC workstation design problems the assignment is well defined by existing constraints, and a conceptual phase might not be needed. The possibilities to effect existing boundaries and technical solutions might be limited and the conceptual phase could be omitted in the process [11]. However, this must come with a warning, as the abstraction of the problem can lead to innovative solutions if time and effort are put into the process. Use of manual and automation constraints is a method to ensure that impossible solutions are excluded in the time-consuming simulation phase. Exactly what is included in these constraints differs between industries, between companies in the same industry as well as between cases in the same company. Manual constraints can differ concerning use of lifting equipment that might enable humans to handle heavy weights in a manual station. Automation constraints differ depending on the production strategy, whether it is a mass production or an engineerto-order production strategy. In mass production, automation barriers can be overcome through advanced sensors that might be economically feasible, as the cost is distributed across a large number of products, while this kind of investment is more difficult to motivate in an engineer-to-order environment. Thus these constraints have to be defined individually for the workstation design problem investigated. In the embodiment design phase the six HIRC design variables were defined: industrial robot variant, industrial robot position, industrial robot gripper design, material position, workstation equipment position and anthropometric database. These were defined through earlier cases and simulations in the HIRC demonstrator software and in research group discussions. The selection of values of these variables is a delicate issue; more feasible values for each variable increase the possibility to find an optimal result early but also require much longer simulation times. Fewer values result in faster simulation but with potentially more simulation loops. Selection of industrial robot variant is a good example of this. It could actually be practically difficult to consider all potential industrial robots available on the market. A company-specific constraint on robot supplier, as mentioned in section 4.1, could actually increase the possibility to find a suitable solution, since the size of the available solution space gets more manageable. The industrial robot gripper design has to be considered; thus the position of the object to handle relative to the industrial robot arm is of great importance to robotic operation time. The workstation equipment position category includes positions of a wide variety that influence the workstation layout. Any designer can also define new design variables in a HIRC design problem if needed. Every design variable has to be analysed in the


planning phase for each workstation as described above. A few of the variables might then be defined as constraints with a fixed value, while others will become design variables. 5.3. Conclusion The HIRC design method presented can be used for design of collaborative workstations before making investment decisions. It incorporates a HIRC simulation demonstrator, which is still under development, into the Pahl and Beitz engineering design framework to cover the process from identifying a potential HIRC station to a finalised design including documentation. This end result is not necessarily a HIRC workstation design; the evaluation of the software might conclude that a fully manual or a fully automatic solution is the best one in the case investigated. The use of the proposed method might enable a sound and clear workstation design decision to be made early in the production development process. Acknowledgements The research work was funded by the Swedish Knowledge Foundation (for the INNOFACTURE Research School), Scania and Mälardalen University. The research is also supported by the research project Virtual Verification of Human Robot Collaboration founded by the Swedish Governmental Agency for Innovation Systems (VINNOVA). The research was conducted in the context of the XPRES research and education environment at Mälardalen University. References [1] Ore F. Human − industrial robot collaboration: Simulation, visualisation and optimisation of future assembly workstations, Lic. thesis, Västerås, Sweden, Mälardalen University, 2015. [2] Krüger J, Lien TK and Verl A. Cooperation of human and machines in assembly lines, Keynote paper, CIRP Annals – Manufacturing Technology, 2009, Vol. 58, pp. 628-646. [3] Krüger J, Nickolay B, Heyer P and Seliger G. Image based 3D Surveillance for flexible Man-Robot-Cooperation, CIRP Annals – Manufacturing Technology, 2005, Vol. 54, pp. 19-22. [4] Helms E, Schraft RD and Hagele M. rob@work: Robot assistant in industrial environments, in Proceedings of the 11th IEEE International Workshop on Robot and Human Interactive Communication, 2002, pp. 399-404. [5] Murphy CA and Perera T. The definition of simulation and its role within an aerospace company, Simulation Practice and Theory, 2002, Vol. 9, pp. 273-291. [6] Klingstam P and Gullander P. Overview of simulation tools for computer-aided production engineering, Computers in Industry, 1999, Vol. 38, pp. 173-186. [7] Mourtzis D, Papakostas N, Mavrikios D, Makris S and Alexopoulos K. The role of simulation in digital manufacturing: applications and outlook, International Journal of Computer Integrated Manufacturing, 2015, Vol. 28, pp. 3-24. [8] Ore F, Hanson L, Delfs N and Wiktorsson M. Human industrial robot collaboration - development and application of simulation software, International Journal of Human Factors Modelling and Simulation, 2015, Vol. 5, pp. 164-185. [9] Khalid O, Caliskan D, Ore F and Hanson L. Simulation and evaluation of industrial applications of Human-Industrial Robot Collaboration cases, paper presented at the Nordic Ergonomics Society 47th Annual Conference, Lillehammer, Norway, 2015. [10] Le Masson P and Weil B. Design theories as languages of the unknown: insights from the German roots of systematic design (1840–1960), Research in Engineering Design, 2013, Vol. 24, pp. 105-126. [11] Pahl G, Beitz W, Feldhusen J and Grote KH, Engineering Design: A Systematic Approach. London, UK: Springer-Verlag 2007. [12] Stauffer L and Pawar T. A comparison of Systematic Design and Design for Six Sigma, in Proceedings of ICED 2007, the 16th International Conference on Engineering Design, 2007. [13] Ulrich K and Eppinger S, Product design and development. New York, NY: McGraw-Hill, Inc, 1995. [14] Hubka V and Eder WE, Theory of Technical Systems: A Total Concept Theory for Engineering Systems. Berlin, Germany: Springer-Verlag, 1988. [15] Pugh S, Total Design: Integrated Methods for Successful Product Engineering. Harlow, UK: Addison-Wesley, 1991. [16] Wu B, Manufacturing Systems Design and Analysis, 2 ed. London: Chapman & Hall, 1994. [17] Bennett D, Production Systems Design. London, UK: Butterworth-Heinemann, 1986. [18] Wiktorsson M. Performance Assessment of Assembly Systems: Linking strategy to Analysis in Early Stage Design of Large Assembly Systems, PhD thesis, Stockholm, Sweden, Royal Institute of Technology, 2000. [19] Bellgran M. Systematic Design of Assembly Systems – Preconditions and Design Process Planning, PhD thesis, Linköping, Sweden, Linköping University, 1998. [20] Pahl G and Beitz W, Konstruktionslehre. Handbuch für Studium und Praxis. Berlin, Germany: Springer, 1977.

11

12


[21] Pini F, Leali F and Ansaloni M. A systematic approach to the engineering design of a HRC workcell for bio-medical product assembly, in 2015 IEEE 20th Conference on Emeerging Technologies & Factory Automation (ETFA),2015, pp. 1-8. [21] Chen F, Sekiyama K, Huang J, Sun B, Sasaki H and Fukuda T. An assembly strategy scheduling method for human and robot coordinated cell manufacturing, International Journal of Intelligent Computing and Cybernetics, 2011, Vol. 4, pp. 487-510. [22] Tsarouchi P, Matthaiakis A-S, Makris S and Chryssolouris G. On a human-robot collaboration in an assembly cell, International Journal of Computer Integrated Manufacturing, 2016, pp. 1-10. [23] Bobka P, Germann T, Heyn JK, Gerbers R, Dietrich F and Dröder K. Simulation Platform to Investigate Safe Operation of Human-Robot Collaboration Systems, Procedia CIRP, 2016, Vol. 44, pp. 187-192. [24] Maurice P, Padois V, Measson Y and Bidaud P. A digital human tool for guiding the ergonomic design of collaborative robots, paper presented at the 4th International Digital Human Modeling Symposium (DHM2016), 2016. [25] Tsarouchi P, Makris S, Michalos G, Matthaiakis A-S, Chatzigeorgiou X, Athanasatos A, Stefos M, Aivaliotis P and Chryssolouris G. ROS Based Coordination of Human Robot Cooperative Assembly Tasks-An Industrial Case Study, Procedia CIRP, 2015, Vol. 37, pp. 254-259. [26] Mårdberg P, Carlson JS, Bohlin R, Delfs N, Gustafsson S, Högberg D and Hanson L. Using a formal high-level language and an automated manikin to automatically generate assembly instructions, International Journal of Human Factors Modeling and Simulation, 2014, Vol. 4, pp. 233-249. [27] McAtamney L and Corlett EN. RULA: a survey method for the investigation of work-related upper limb disorders, Applied Ergonomics, 1993, Vol. 24, pp. 91-99. [28] Ore F, Hanson L, Wiktorsson M and Eriksson Y. Automation constraints in human–industrial robot collaborative workstation design, paper presented at the 7th International Swedish Production Symposium, Lund, Sweden, 2016. [29] Stanton NA. Hierarchical task analysis: Developments, applications, and extensions, Applied Ergonomics, 2006, Vol. 37, pp. 55-79. [30] Fasth Å, Provost J, Fabian M, Stahre J and Lennartson B. From Task Allocation Towards Resource Allocation when Optimising Assembly Systems, Procedia CIRP, Vol. 3, 45th CIRP Conference on Manufacturing Systems 2012, 2012, pp. 400-405.