Methodological Approach for the Design of a Complex Inclusive Human-Machine System
Lorenzo Sabattini, Valeria Villani, Julia N. Czerniak, Alexander Mertens, Cesare Fantuzzi
MMethodological Approach for the Designof a Complex Inclusive Human-Machine System
Lorenzo Sabattini , Valeria Villani , Julia N. Czerniak , Alexander Mertens and Cesare Fantuzzi Abstract — Modern industrial automatic machines androbotic cells are equipped with highly complex human-machineinterfaces (HMIs) that often prevent human operators froman effective use of the automatic systems. In particular, thisapplies to vulnerable users, such as those with low experienceor education level, the elderly and the disabled. To tackle thisissue, it becomes necessary to design user-oriented HMIs, whichadapt to the capabilities and skills of users, thus compensatingtheir limitations and taking full advantage of their knowledge.In this paper, we propose a methodological approach to thedesign of complex adaptive human-machine systems that mightbe inclusive of all users, in particular the vulnerable ones.The proposed approach takes into account both the technicalrequirements and the requirements for ethical, legal and socialimplications (ELSI) for the design of automatic systems. Thetechnical requirements derive from a thorough analysis of threeuse cases taken from the European project INCLUSIVE. Toachieve the ELSI requirements, the MEESTAR approach iscombined with the specific legal issues for occupational systemsand requirements of the target users.
I. I
NTRODUCTION
Advances in technology in modern industrial settings haveled to the introduction of extremely complex automaticmachines and robotic cells. Despite such a massive introduc-tion of advanced technological solution, the role of humanoperators in this context is still focal, since they are responsi-ble for controlling and supervising manufacturing activitiesand the desired flexible production. Nevertheless, this newtechnological scenario is not favorable to human operatorsthemselves: indeed, the complexity of modern manufacturingplants is reflected in an increased complexity of the ac-companying human-machine interfaces (HMIs), which allowthe user to operate the machine, observe the system statusand, if necessary, intervene in the process [1], [2]. Theincrease in complexity of modern industrial HMIs can still betackled by the most experienced human operators, who caninteract efficiently with the machine only at the expensesof an unsustainably increased mental workload and stress.However, in the worst condition, vulnerable workers, suchas those with low experience or education level, the elderlyand the disabled, can barely sustain such an interaction inan effective manner.To tackle this issue, it is needed to make use of ananthropocentric approach that reverses the paradigm from L. Sabattini, V. Villani and C. Fantuzzi are with the Department ofSciences and Methods for Engineering (DISMI), University of Modenaand Reggio Emilia, Reggio Emilia, Italy { lorenzo.sabattini,valeria.villani, cesare.fantuzzi } @unimore.it J. N. Czerniak and A. Mertens are with the Institute of IndustrialEngineering and Ergonomics, RWTH Aachen University, Aachen, Germany { j.czerniak, a.mertens } @iaw.rwth-aachen.de the current belief that ”the human learns how the machineworks” to the future scenario in which ”the machine adaptsto the human capability” accommodating to her/his own timeand features [3]. This is realized by adaptively simplifyingthe HMI based on the user’s features and complementingher/his cognitive capabilities by advanced sensing and higherprecision of machines. Following such approach, it would bepossible to create an inclusive [4], [5] and flexible workingenvironment for any kind of operator, taking into accountmultiple cultural backgrounds, skills, age and different abil-ities. developing a methodology for the design of adaptivehuman-centered HMIs for industrial machines and robots.HMIs typically used for supervising industrial processesdo not provide any possibility of controlling the amount ofdisplayed information, or its form. Hence, while the humanoperator is flexible and adaptable, the system is not. Inparticular, the control systems applied to industrial processestypically respond in a specified way, without regard as towhether the flow of information is low or extremely high,or the level of expertise of the user is good or bad [6]. Thehuman operator is then typically the only element that needsto adapt her/his behavior based on the situation. Namely, theoperator needs to be sufficiently flexible, to be able to copeboth with common activities and unpredictable situations,such as in the presence of dangers. This can cause significantdifficulties for the operators, in particular considering the factthat the amount of monitored data that come from modernproduction processes is constantly increasing, and controlsystems are becoming increasingly complex [1], [6], [7].To overcome this issue, the concept of context-dependentautomation, also known as adaptive automation, has beenintroduced [8], [9]. Generally speaking, context awarenessis the ability for a system to sense, interpret, respond andact based on the context [10]. Based on this paradigm, thelevel of automation of a system is designed to be variable,depending on situational demands during operational use.Along similar lines, the idea of adaptive user interfaceshas been developed, which consist in changing how theinformation is presented, in such a way that only the relevantpieces of information are provided to the operator, based onthe context. Examples of adaptive user interfaces have beendeveloped considering different application domains, such asautomotive [11], [12], aeronautics [13] and smartphones andhand-held devices [14]. However, to the best of the authors’knowledge, only a few pioneering examples have been pre-liminary presented regarding HMIs for complex industrialsystems [6], [9]. Specifically, [6] described a preliminaryconcept of architecture for an HMI that adapts the presen- a r X i v : . [ c s . H C ] J un ation of information based on the operator responsiveness.Profiling of the operators is considered in [9], and the HMIselectively presents information based on the profile of thecurrent user.Going beyond this state of the art, the European projectINCLUSIVE aims at developing a smart interaction systemthat adapts the information load of the HMI and the automa-tion capability of the machine to the physical, sensorial andcognitive capabilities of workers [15]. In particular, the finalgoal is to provide technological solutions for compensatingworkers’ limitations (e.g. due to age or inexperience), whiletaking full advantage of their knowledge. Three groups ofoperators are considered, namely elder, disabled, and inex-perienced operators, since they are believed to be the mostvulnerable ones in the interaction with complex automaticsystems, as discussed in Sec. II.Three main pillars constitute the INCLUSIVE system[15]. The first pillar relates to the measurement of humancapabilities: the system will measure the human capability ofunderstanding the logical organization of information and thecognitive burden the operator can sustain (automatic humanprofiling). The second pillar consists in the adaptation ofinterfaces to human capabilities: the system will adapt theorganization of the information, the means of interaction, andthe automation task that are accessible by the user dependingon her/his measured capabilities. Finally, the third pillar isabout teaching and training for unskilled users: the systemwill be able to teach the correct way to interact with themachine to the unskilled users, exploiting also simulation invirtual and augmented environment.In this paper, we propose a set of methodological recom-mendations for the design of an adaptive human-machinesystem that is inclusive for all users. In particular, we derivethe technical requirements that a complex human-machinesystem, such as the one considered in INCLUSIVE, shouldfulfill in order to allow also vulnerable users to access it.Such requirements are defined starting from the analysis ofthe industrial use cases of INCLUSIVE, but have generalvalidity. In particular, the main issues related to state ofthe art solutions in terms of HMI are highlighted, referringexplicitly to representative target scenarios. From the analysisof the use cases, a set of users’ needs is defined. Specifically,users’ needs describe the technical issues and difficultiesthat operators typically encounter with the currently availabletechnological solutions. Users’ needs are then abstracted, todefine the technical system requirements. These are generaltechnical methodological guidelines that should be consid-ered in the design of any complex human-machine system,in order to make it accessible also to vulnerable users.Moreover, we carry out an analysis of the different ethical,social and legal implications (ELSI) of such a system toprotect the user against harm and disadvantages. Based onthe MEESTAR approach [16], which is an instrument foridentifying ethical problems, we develop an ELSI conceptand test its appropriateness in a possible operative scenario.Then, we derive some design recommendations in terms ofELSI requirements for the development of smart interaction systems for automated production machines. The aim isoffering fair requirements, independent of individual skillsand capabilities.II. D ESCRIPTION OF THE CONSIDERED USE CASES
To derive methodological considerations that have generalvalidity it is important to start from real use cases that depictthe scenario of human-machine systems currently utilized inindustrial environments. To this end, we consider, as a casestudy, the industrial use cases addressed in the INCLUSIVEproject, since they are representative of a wide area of interestfor industry in Europe:
Use case 1 : machinery for small companies, typically runby elderly owners;
Use case 2 : automation solutions made for developingcountries;
Use case 3 : industrial plants made by a big company.Specifically, the first use case refers to machinery used forwoodworking in artisans’ shops. The second one considersa robotic solution to be applied in a company located in adeveloping country, where operations are mostly performedmanually. In particular, the considered robotic solution is forpanel bending. Finally, the third use case refers to a bottlingcompany and, in particular, a labelling unit is considered.Such use cases have been chosen since they addressdifferent categories of most vulnerable users, namely elderly,disabled and low experienced. Specifically, by elderly weconsider those people in the last years of their work life.Generally, these workers have a large experience in thetraditional industrial processes, but are not familiar withmodern computerized devices and, then, have difficultiesin utilizing modern automatic machines that come withcomplex HMIs. As regards people with physical impairmentand limited cognitive abilities, such limitations introduce aswell difficulties in the use of complex automatic machines.Finally, by inexperienced we refer to people with low levelof education, limited expertise in the use of automaticmachines and/or computerized HMI, and lack of experiencein industrial processes.For each use case, a specific working scenario is analyzedin order to derive what are the concrete limitations of cur-rently implemented solutions. These activities were selectedby the corresponding industrial partners of the INCLUSIVEconsortium, since they require unavoidable interaction of theuser with the machine and are representative of the mostfrequent operations with automatic machines. Specifically,for the first use case we focus on the activities related totuning of the machine, to make it ready for woodworking(tuning of the tools warehouse, tuning of the worktablearea components) and routine maintenance procedures. Forthe second use case, we consider the standard activitiesperformed by a user for bending a part, and replacingmalfunctioning tools. The working scenario for the third usecase refers to the fault recovery procedure, performed in jogmode, for misalignment of the neck ring label of bottlesand the changeover of the printing format, required at theeginning of a working day or when a new bottle or label isproduced on the line.III. A
NALYSIS OF THE PROBLEMS OF CURRENT
HMI S For each of the working scenarios, we analyzed howinteraction is currently carried out, aiming at finding pitfallswhich should be corrected in an inclusive system.
A. Use case 1
The first limitation in the current implementation of thehuman-machine interaction lies in the fact that there isa clear lack of guided procedures assisting the user. Infact, the user is currently barely supported by the interface:only simple alarms are displayed, which describe what thecurrent problem is, but not how to solve it. Moreover, asregards the setup of the tools change, there is a misalignmentbetween the equipment in the physical store (i.e., the toolson board the machine) and that in the virtual one shownin the HMI (i.e., the tools that the HMI displays as onboard the machine), since the virtual store does not updateautomatically when a change in the physical one is made.As a consequence, currently the operator must pay attentionto avoid mistakes that could jeopardize the operation of themachine: clearly, this activity is time-consuming and proneto errors.This consideration applies also to the setup of the workingarea. Indeed, currently the interface supports the operatoronly by displaying, in a picture, the position of the com-ponents. It is up to the operator to manually move thevarious components in the correct position. This lack ofintuitiveness and assistance results in an additional decreaseof efficiency and raises problems related to the constantneed to consult the operator’s manual, thus stopping normaloperations to solve routine issues. However, since the manualis typically not stored close to the machine and is notorganized with a clear focus on troubleshooting, it is rarelyused by the operators, who end up to directly contact theassistance service to solve routine issues. In some othercases, they perform some tasks following some unofficialshortcuts rather than the official procedures recommendedin the manual. Moreover, given the lack of guidance, oftenerrors of inexperienced operators severely compromise theoperation of the machine.
B. Use case 2
With respect to the second use case, the main problemof the current HMI lies in the fact that these robotic cellscan be used only by highly skilled personnel. In particular,background education in mechanical or electrical fields isnecessary, since operators need to have significant codingskills both to program the system, and to be able to recoverfrom problems that could arise during normal operations,also for simple cases, such as photocell malfunction. The useof the system by unskilled operators usually causes severalproblems since they often choose the wrong tool to performthe bending operations, or the wrong material thickness,thus making bending not possible, or wrong settings in the definition of the air pressure, that thus leads to incorrectbending operations. Moreover, current HMIs are based ontouch screens and standard computers, and they cannot beutilized by people with disabilities of the upper limbs, orby blind people, effectively. Further, as in the previous usecase, no guided procedure is available, besides the manual:hence, only operators with a long experience are able tosolve problems. Although several choices need to be madefor setting up the system (e.g. the correct angle to be usedfor bending a certain part), commonly adopted solutionsexist, but they are known only by expert operators. Also,the operator needs to decide what parameters need to bechanged, and then see what the result will be: again, thisoperation is mainly based on the operator’s experience.
C. Use case 3
As regards the last use case, one of the biggest issuesis related, also in this case, to the fact that the use of thesystem by untrained users is impractical. Operators need aspecific training phase, before being able to interact withthe machine. In particular, during the first uses, operatorsperceive the interaction with the system as uncomfortable. Inthese conditions, it was reported that operators feel afraid ofdamaging the system, the machine or the product, especiallyif a trainer or a supervisor is close by: indeed, although thesepeople are trying to help or prevent disasters, the employeeis stressed by this situation even more. Moreover, anothersource of stress is the fact that operators do not receiveany feedback or acknowledgement of performed activities,to help them to understand if they are doing well.Also in this case, inexperienced operators need the manualto check for every possible fault cause and how to correctthem. Despite of this, it still happens that often wrongoperations are performed, or operations are not correctlyperformed according to the manual, and, in particular, oftenthe wrong operational mode is selected, e.g. semiautomaticor manual instead of jog mode. All these issues appear, inparticular, for operators that are new to machinery or for loweducated people.IV. D
EFINITION OF USERS ’ NEEDS
The users’ needs have been identified from the aboveanalysis of the problems of current interaction systems.The first category of users’ needs refers to the inclusionof all users in complex human-machine systems. The systemshould be effectively usable by inexperienced operators,by operators with different age, level of work experience,namely novice users and expert operators, and education,and those with physical impairment. Specifically, the pres-ence of an easily accessible guidance, which might exploitaugmented reality for step-by-step guided procedures, couldbe a substantial advantage for unskilled operators, in orderto make problem solving tasks accessible also to them. Inthis regard, programming by code writing, which is currentlyrequired in the scenario of the second use case, shouldnot be necessary. As regards physical impairments, differentdisabilities might be typical, depending on the applicationcenario: as an example, in the case of woodworking ma-chines, missing fingers have been reported as a typicalimpairment.Thus, a second group of users’ needs rises: the organiza-tion of information should be user-oriented. This implies that,on the one side, procedures should adapt to the operator’sskills, thus being sufficiently clear for unskilled operatorsand not too long-winded for the skilled operators. On theother side, the system should guide the operator during or-dinary operations, such as setup or maintenance. A teachingmodule could be implemented, to suggest unskilled operatorscommon practice solutions. As a consequence, specific priortraining and studying the manual should not be necessary.Despite of this, it should be possible to perform operationsin the correct sequence, according to the manual, by meansof proper suggestions suitably provided by the HMI. Thisshould be possible also for tunable procedures, where thesystem should suggest the operator what parameters needto be changed, based on the desired result. A solution forunskilled operators could be to provide suggestions on whatparameters need to be changed, knowing how they influencethe achieved result.These users’ needs lead to the consideration that humanfactors must be prioritized. Indeed, the system should beperceived as comfortable for all the users and the stress levelduring the use of the system should be low. In order for thisto be achieved, the intervention of supervisors for assistingthe operators should be avoided and operators should feelconfident when using the system alone.As a consequence of such an anthropocentric approach, theoperator’s performance should be automatically enhanced, inthe sense that the operators should be enabled to perform thecorrect actions and choices. The number of errors shouldbe reduced, while the execution time should be improved.Specifically, the correct operational mode and the correctvalue for critical parameters should be automatically se-lected. Also, the choice of wrong options should be preventedand the HMI should depict the actual current equipment andstate of the machine.Finally, some advanced technological solutions shouldbe implemented to allow a smoother interaction with themachine. Specifically, hands-free interaction, such as speechrecognition and synthesis, should be possible to enable theoperators to interact with the machine when wearing glovesor protection equipment. Additionally, portable interfaces,such as wearable devices and augmented reality, should beavailable, to guide the operators in the working area.V. T
ECHNICAL REQUIREMENTS
Based on the description of the use cases and of theidentified user issues, the following system requirementsare derived. They describe how an adaptive human-machinesystem should be implemented in order to be inclusive for allusers, and in particular elderly, disabled and low experiencedusers:
T-R1
The interface adapts to the level of skills of theoperator. xy z
Fig. 1. MEESTAR model: x -axis: dimensions of ethical evaluation; y -axis:stages of ethical evaluation; z -axis: levels of ethical evaluation. T-R2
The system can be used by low educated operators.
T-R3
The system can be used by physically and cognitivelyimpaired operators.
T-R4
The system can be used by people with low computerskills.
T-R5
The system enforces the correct procedures.
T-R6
The operator feels satisfied from the interaction expe-rience.
T-R7
Interaction with the system generates a low level ofstress for the operators.VI. E
THICAL , SOCIAL AND LEGAL ASPECTS
The introduction of a system that processes sensitivepersonal data to disclose barriers of human capabilities,requires that also ethical, legal and social requirements haveto be taken into account to protect the user against harm anddisadvantages. However, evaluating ethical, social and legalimplications (ELSI) represents a specific challenge. In thispaper we propose to deal with ELSI aspects by a diverseapproach, namely the MEESTAR model, which originallywas developed for evaluating socio-technical arrangementsin the field of age appropriate assisting systems [16]. It is ananalytical instrument which guides the process of reflectingon the use of technology. The model aims at identifyingethically problematic effects in a structured way and, onthat basis, develop appropriate solutions. The model focuseson negative effects, requiring that the system causes littleor no harm to the user. The first step of the MEESTARanalysis is to identify relevant ethical dimensions for theparticular scenario. Thus, the aim of this approach is to finda basis for ethical, social and legal aspects, according tothe intention of implementing sensors for measuring humancapabilities and tracking individual health data. Furthermore,legal requirements given by the European Union (EU) areconsidered , and finally responsibility for needs of vulnerabletarget users is taken into account.Working with MEESTAR involves the systematic con-sideration of three axes, as shown in Fig. 1. The x -axisconsists of seven ethical dimensions: care, autonomy, safety, In this paper we consider only EU legislation.ig. 2. The intersection of ethical, legal and social implications (ELSI)define the requirements for the considered inclusive human-machine system. justice, privacy, participation and self-conception. The y -axisdescribes stages of ethical evaluation, allocating problemsamong four levels of ethical sensitivity. The z -axis providesthree points of view (individual, organizational, social).The legal issues regard mainly data protection, safetyand health at work, and product requirements. The maindirectives in the context of production machines are the Ma-chinery Directive 2006/42/EC about construction of safety-related products, and the Council Directive 89/391/EEC onthe introduction of measures to encourage improvements inthe safety and health of workers at work.The MEESTAR dimensions show several intersectionswith legal requirements and target users, as shown in Fig. 2: • caring for users with different limitations in skills andcapabilities, • giving these users possibility for an autonomous inter-action with automated production systems, • fulfilling standards for safety and justice, by addressingemployers corporate duties by law, • sensitively approaching the employees right to privacyaccording to legal requirements, by treating personaldata with dignity and respect.Thus, the following technical aspects that need to be takeninto account can be derived: i) occupational health, ii) occu-pational safety, iii) data protection, iv) ergonomic workplacedesign, v) equal opportunities and vi) reintegration. Specif-ically, occupational safety and health is an interdisciplinaryfield, concerning safety and health of a working person inan occupational system to prevent him/her from workinghazards [17] in accordance with the MEESTAR dimensions safety and justice . Also ergonomic workplace design, as asubtask of occupational safety and health promotion, belongsto this category. Under EU law, data can only be processedunder strict conditions, because everybody has a right tothe protection of personal data [18], which corresponds toMEESTAR dimension privacy . In the perspective of targetusers, who have special characteristics and therefore differ-ences in perception, cognition and motor skills, an equaltreatment and integration into working processes is required.Thus, care about their capabilities and autonomous use ofautomated machines are the main topics in this case.VII. A SSESSMENT OF THE
ELSI
CONCEPT
To assess the discussed dimensions of the ELSI concept,a questionnaire was developed to investigate the appropriate-
TABLE IELSI
CONCEPT : POTENTIAL OF IMPROVEMENT AND RISKS . Dimension Potential of improvement Risks
Occupationalhealth - Health detection - HMI can adapt to tasks hard to accomplish - Avoid injuries or dangerous operation - Qualified personal improvement - Less mistakes in production -‐
Intrusive measures -‐
Strain through dynamic changing processes -‐
Put more strain to the worker instead of helping him due to production capacity needsOccupationalsafety - Prevention from dangerous procedures through HMI - Higher level of attention avoiding dangerous operations - Improvement of working conditions - Less strain results in more safety - Limitation of potential injuries - Predict and estimate risk & strain becomes complex operation - System reassures user - Obstacles in operation (for example risk of stumble in cablesDataprotection - Record all parameters and control the personal machine conformity - Easy adaption according to person’s capacities - Physical improvement by improvement of transferring tools - Risk about sharing the related information with the personal machine - Personal background can be responsible for strain - Data confidentiality must be guaranteed - Operator could feel monitored, controlled ( à measurement has to be acknowledged)Ergonomicworkplacedesign - Also workplace adaption according to measured strain possible - Rigid postures - Incorrect adaption could result in higher strain levels - Obstacles in safe movement - Dependency of ergonomic design on strainEqualopportunities - Strain level could be indicator for continuous harassment - Increase self-confidence - Increase expectation of working quality - Adaption to disparate capabilities - HMI is not usable equally for users with different requirements - False impression of safety could occur Reintegration - HMI can report, if an operator loses ability to accomplish a task - Potentiation of skills - Compensation of deficits - Monitoring to strain caused by illness - Operator could feel monitored & controlled about capabilities, resulting in higher strain levels ness of the identified dimensions in the considered scenario,namely that of an inclusive complex human-machine systemaccessible to special user groups, with special needs andrequirements. To make the participation in the questionnairemore effective, we considered a specific working contextwhere affective computing is applied to an industrial human-machine system, thus measuring operator’s mental workload,stress and induced anxiety by recording some physiologicalsignals. Specifically, the questionnaire included questionsregarding the following scenario: ”
The working machinesare equipped with sensors that are able to track strainof a working person by real-time measurement of his/herphysiological parameters, e.g. heart rate, blood pressure, etc.If the measured strain indicators are too high, the human-machine-interface adapts to the situation resulting in a lowerstress level. ”The questionnaire was distributed to all members of theINCLUSIVE consortium, to consider all relevant stakehold-ers that are affected. Seven partners participated in the studyand participants were employed at companies in the follow-ing sectors: IT, technology transfer, industrial automation,white goods, packaging and bottling. Each participant in thestudy was asked whether a potential of improvement/risksin measuring strain of a working person is measured ac-cording to each of the dimensions of the ELSI concept,namely occupational health, occupational safety, data pro-ection, ergonomic workplace design, equal opportunitiesand reintegration. Table I lists in detail all the potential ofimprovement and risks mentioned by the participants in thestudy. In particular, when designing the HMI, it has to betaken into account that the complexity resulting from theadaptive HMI behavior prevents inducing strain itself. Inaddition, the system must implement effective anonymizationof personal user data; otherwise, there would be the risk thatperformance assessment, for instance, leads to a terminationof employment. Moreover, the system should ensure thatnobody is discriminated. According to respondents’ answers,the supporting system should also ensure that the users haveto respect safety regulations. Here, the system should meetrelevant safety criteria and, if false impression of securityoccurs, call her/his attention. The measuring system shouldalso take into account that the user is not distracted whileworking and that there is not a risk of stumbling. Accordingto doubts of participants, the system should in no case causeinjury to health by means of inductive measuring technology.
A. ELSI requirements
The findings reported in Table I allow us to derive theELSI requirements, which have general validity and thusapply to any user-centred human-machine system that re-lies on affective computing for including vulnerable users.Specifically, the derived design recommendations for ethical,social and legal aspects are the following:
ELSI-R1
The system prevents inducing strain itself.
ELSI-R2
The system considers anonymized personal data.
ELSI-R3
The system uses collected data not for any disad-vantage for the employer.
ELSI-R4
The system depicts relevant user requirements andprevents discrimination.
ELSI-R5
The system meets all relevant safety criteria.
ELSI-R6
The system does not distract the operator.
ELSI-R7
The system does not cause injuries by means ofinductive measuring technology.VIII. C
ONCLUSION
In this paper we presented a methodological approach tothe design of complex human-machine systems that adaptto the operators skills and capabilities, complementing theirlimitations, while taking full advantage of their knowledge.Specifically, the proposed approach aims at guiding in thedesign of HMIs that can be effectively used by vulnerableoperators, such as those with low experience or educationlevel, the elderly and the disabled. To this end, we defineda set of technical requirements and requirements related toethical, legal and social implications. The technical require-ments were derived from the analysis of the industrial usecases considered in the European project INCLUSIVE andthey abstract what should be fulfilled in order to allow alsovulnerable users to access a complex automatic machineor robotic cell were derived. As regards the ethical, legaland social requirements, they were derived combining theMEESTAR approach with the specific legal issues for occu-pational systems and requirements of the target users. The validity of the such requirements was then validated in thecontext of the INCLUSIVE project.A
CKNOWLEDGEMENT
The research is carried out within the ”Smart and adap-tive interfaces for INCLUSIVE work environment” project,funded by the European Union’s Horizon 2020 Research andInnovation Programme under grant agreement N723373. Theauthors would like to thank the industrial partners responsiblefor the use cases for providing a description of the use casesand selected working scenarios.R
EFERENCES[1] T. Skripcak, P. Tanuska, U. Konrad, and N. Schmeisser, “Toward non-conventional human-machine interfaces for supervisory plant processmonitoring,”
IEEE Trans. Human-Machine Systems , vol. 43, no. 5, pp.437–450, Sep. 2013.[2] F. Nachreiner, P. Nickel, and I. Meyer, “Human factors in processcontrol systems: The design of human–machine interfaces,”
SafetyScience , vol. 44, pp. 5–26, 2006.[3] D. A. Norman,
The design of everyday things: Revised and expandededition . Basic books, 2013.[4] J. Abascal and C. Nicolle, “Moving towards inclusive design guide-lines for socially and ethically aware HCI,”
Intercating with Comput-ers , vol. 17, pp. 484–505, 2005.[5] C. Stephanidis and A. Savidis, “Universal access in the informationsociety: Methods, tools, and interaction technologies,”
Universal Ac-cess in the Information Society , vol. 1, no. 1, pp. 40–55, 2001.[6] G. Viano, A. Parodi, J. Alty, and et al., “Adaptive user interface forprocess control based on multi-agent approach,” in
Proc. WorkingConf. Advanced Visual Interfaces , ser. AVI ’00. New York, NY,USA: ACM, 2000, pp. 201–204.[7] E. Flasp¨oler et al., “The human machine interface as an emergingrisk,” European Agency for Safety and Health at Work, Tech. Rep.,2010.[8] R. Parasuraman, T. Sheridan, and C. Wickens, “A model for types andlevels of human interaction with automation,,”
IEEE Trans. Systems,Man, and Cybernetics - Part A: Systems and Humans , vol. 30, no. 3,pp. 286–297, May 2000.[9] A. Lee and J. Martinez Lastra, “Enhancement of industrial moni-toring systems by utilizing context awareness,” in
IEEE Int. Multi-Disciplinary Conf. Cognitive Methods in Situation Awareness andDecision Support (CogSIMA 2013) , 2013.[10] A. K. Dey, “Understanding and using context,”
J. Personal andUbiquitous Computing , vol. 5, no. 1, pp. 4–7, Feb. 2001.[11] H. Sharma, R. Kuvedu-Libla, and A. Ramani, “ConFra: A contextaware human machine interface framework for in-vehicle infotainmentapplications,” in
Int. MultiConf. Engineers and Computer Scientists(IMECS 2008) , 2008, pp. 19–21.[12] S. Garzon and M. Poguntke, “The personal adaptive in-car HMI:Integration of external applications for personalized use,” in
Advancesin User Modeling , S. B. Heidelberg, Ed., 2012, pp. 35–46.[13] T. Inagaki, “Situation-adaptive autonomy: Dynamic trading of author-ity between human and automation,” in
Proc. Human Factors andErgonomics Society Annual Meeting , vol. 44, Jul. 2000, pp. 13–16.[14] T. Gu, H. Pung, and D. Zhang, “A middleware for building context-aware mobile services,” in
IEEE Vehicular Technology Conf. (VTC2004) , 2004, pp. 2656–2660.[15] V. Villani, L. Sabattini, J. N. Czerniak, A. Mertens, B. Vogel-Heuser,and C. Fantuzzi, “Towards modern inclusive factories: A methodologyfor the development of smart adaptive human-machine interfaces,” in . IEEE, 2017.[16] A. Manzeschke, K. Weber, E. Rother, and H. Fangerau, “Ethicalquestions in the area of age appropriate assisting systems,”
GermanFederal Ministry of Education and Research, VDI/VDE Innovation+Technik GmbH , 2015.[17] P. Hughes and E. Ferrett,
Introduction to health and safety at work .Routledge, 2011.[18] R. H. Weber, “Internet of things–new security and privacy challenges,”