Chapter 9.1 Perspectives on Designing Human Interfaces for Automated Systems Anil Mital University of Cincinnati, Cincinnati, Ohio Arunkumar Pennathur University of Texas at El Paso, El Paso, Texas 1.1 INTRODUCTION 1.1.1 Importance and Relevance of Human Factors Considerations in Manufacturing Systems Design The design and operation of manufacturing systems continue to have great signi®cance in countries with large and moderate manufacturing base, such as the United States, Germany, Japan, South Korea, Taiwan, and Singapore. It was widely believed in the 1980s that complete automation of manufacturing activities through design concepts such as ``lights-out factories,'' would completely eliminate human in¯u- ence from manufacturing, and make manufacturing more productive [1]. However, we now see that com- plete automation of manufacturing activities has not happened, except in a few isolated cases. We see three basic types of manufacturing systems present and emergingÐthe still somewhat prevalent traditional manual manufacturing mode with heavy human in- volvement in physical tasks, the predominant hybrid manufacturing scenario (also referred to traditionally as the mechanical or the semiautomatic systems) with powered machinery sharing tasks with humans, and the few isolated cases of what are called computer- integrated manufacturing (CIM) systems with very little human involvement, primarily in supervisory capacities. Indeed, human operators are playing, and will continue to play, important roles in manufacturing operations [2]. Another important factor that prompts due consid- eration of human factors in a manufacturing system, during its design, is the recent and continuous upward trend in nonfatal occupational injuries that has been observed in the manufacturing industry in the United States [3]. While these injuries may not be as severe and grave as the ones due to accidents such as the Chernobyl Nuclear Reactor accident (the Three Mile Island nuclear accident prompted an upswing in human factors research, especially in nuclear power plants and in process industry settings), the increasing trend in injuries leaves the claim that ``automation'' of manufacturing has resulted in softer jobs for manufac- turing workers questionable. In fact, many manufac- turing researchers and practitioners believe that an increase in severe injuries in manufacturing is primarily due to the automation of simpler tasks, leaving the dicult ones for the humans to perform. This belief is logical as the technology to automate dicult tasks is either unavailable or expensive. The factors discussed suggest that manufacturing systems (our de®nition of a system is broad; a system may thus be a combination of a number of equipment/ machines and/or humans) be designed with human limitations and capabilities in mind, if the system is to be productive, error-free, and safe, and result in 749 Copyright © 2000 Marcel Dekker, Inc. quality goods and services, all vital goals for manufac- turing organizations. 1.1.2 The Human±Machine System Framework for Interface Design Traditionally, system designers have accounted for human limitations and capabilities by considering the human operator as an information processor having sensory and motor capabilities and limitations (Fig. 1). It can be readily seen from Fig. 1 that the key elements to the ecient and error-free functioning of a human±machine system are the provision of infor- mation to human operators in the system, and the provision for control of the system by humans. Displays provide information about the machine or the system to human operators, and controls enable human operators to take actions and change machine or system states (conditions). Operator feedback is obtained through interaction with the controls (tactile sensing, for instance). Thus, in the classical view, human interaction with automation is mediated through displays and controls for a two-way exchange of information. The recent view of the human±machine system, resulting out of advances in computerized informa- tion systems, sees the human operator as a super- visory controller [4] responsible for supervisory functions such as planning, teaching, monitoring, intervening,learning,etc.(Fig.2).Eventhough,in such a view, the human operator has a changed role, displays and controls still provide the fundamental medium for human interaction with the system. Indeed, properly designed displays and controls are fundamental to the ecient and error-free functioning 750 Mital and Pennathur Figure 1 Traditional representation of human interaction with machine. Copyright © 2000 Marcel Dekker, Inc. of manufacturing systems. Ergonomics, which we de®ne as the study of issues involved in the application of technology to an appropriate degree to assist the human element in work and in the workplace, provides recommendations for interface design based on research in human sensory and motor capabilities and limitations. 1.1.3 Scope of This Chapter Even though displays and controls, and their eective design, are fundamental to the ecient and error-free operation of the system, a number of important activ- ities need to be carried out before one can think of displays and controls. These activities stem from the central need to build systems to suit human limita- tions and capabilities. Some of these activities, such as ``user needs analysis,'' are relatively new concepts and form the core of what is called the ``usability engineering approach'' to design. Techniques asso- ciated with other activities, such as task analysis and function allocation between humans and auto- mated equipment, are an integral part of designing ``good'' jobs, and have been in existence for some time. We present some of these techniques and meth- ods. Inherent throughout our presentation is the essence of the ``human-centered interface design approach.'' We ®rst present elements of this approach and con- trast it with the ``system-centered interface design approach.'' It is recommended that this concept of human-centered design guide the designer at both the system, as well as at the nuts-and-bolts, design levels. Displays and controls, the selection, design, and evaluation of which will be the theme for the remainder of the chapter, form a part of aids, equipment, tools, devices, etc., that are necessary for a system to operate satisfactorily. Due to the wide variety of available tech- nologies, and due to the fact that most ergonomics recommendations for the design of displays and con- trols remain the same regardless of the technology used (e.g., recommendations on the design of lettering remain the same whether the recommendation is for a conventional hand-held meter, a visual display unit, or printed material), we provide only general recom- mendations for dierent types of displays and controls, without reference to commercial products and equipment. A few other notes about the scope of this chapter: due to the vast literature available in the area of design of human±machine systems, our emphasis in this chapter is on the breadth of coverage rather than depth in any area. This emphasis is deliberate, and is motivated, in addition, by our intention to provide the reader a taste of the process of design and evaluation of a modern human±machine system. Readers interested in more detail in any one area or technique should refer to our recommended reading list. Also, even though the recommendations and guidelines summarized in this chapter come from research in human±machine settings other than hard- core manufacturing settings, they are equally applic- able to manufacturing systemsÐthe general framework and the speci®c recommendations we have collected and provided in this chapter for design of human±machine systems are applicable across systems. Human Interfaces for Automated Systems 751 Figure 2 The latest notion of human as a supervisory controller. Copyright © 2000 Marcel Dekker, Inc. 1.2 APPROACHES TO DESIGNING SYSTEMS FOR HUMAN±MACHINE INTERFACE 1.2.1 The System-Centered Design Approach The system-centered design approach, as the name sug- gests, analyzes the system currently in use, designs and speci®es the new system based on this analysis, builds and tests the new system, and delivers the system and makes minor changes to the system (Fig. 3). The focus is on the goals of the system and the goals of the orga- nization within which the system is to perform. Designers following this approach fail to consider the users before designing the system. As a result, users of such systems are required to remember too much infor- mation. Also, typically, these systems are intolerant of minor user errors, and are confusing to new users. More often than not, such systems do not provide the functions users want, and force the users to per- form tasks in undesirable ways. New systems designed the system-centered way have also been shown to cause unacceptable changes to the structure and practices in entire organizations [5]. 1.2.2 The Human-Centered Design Approach The human-centered design approach to human± machine interaction, unlike the system-centered approach, puts the human attributes in the system ahead of system goals. In other words, the entire system is built around the user of the systemÐthe human in the system. This approach has been var- iously called the ``usability engineering approach,'' the ``user-centered approach'' or the ``anthropocentric approachtoproductionsystems,''etc.Figure4pro- vides our conception of the human-centered approach to interface design. The ®rst step in this design approach is information collection. Information about user needs, information about user cognitive and mental models, information on task demands, information on the environment in which the users have to perform, information on the existing interface between the human operator (the user of the system) and the machine(s), requirements of the design, etc., are some of the more important variables about which information is collected. This information is then used in the detailed design of the new interface. The design is then evaluated. Prototype development and testing of the prototype are then performed just as in any other design process. User testing and evalua- tion of the prototype, the other important characteris- tic of this design process which calls for input from the end user, is then carried out. This results in new input to the design of the interface, making the entire design process iterative in nature. Even though the human-centered design approach is intended to take human capabilities and limitations into account in system design and make the system usable, there are a number of diculties with this approach. The usability of the system is only as good as its usability goals. Thus, if the input from the users about the usability goals of the system are inappropri- ate, the system will be unusable. One approach to over- come this problem is to include users when setting usability goals; not just when measuring the usability goals. Another common diculty with this approach is the lack of provision to take into account qualitative data for designing and re®ning the design. This is due to the de®ciency inherent in the de®nition of usability which calls for quantitative data to accurately assess the usability of a system. There is also the drawback that this approach is best suited for designing new systems, and that it is not as eective for redesign of existing systems. Despite these limitations, the human-centered design approach merits consideration from designers because it proactively takes the user of the product (displays and controls with which we are concerned, and which make up the interfaces for human±machine interaction, are products) into the system design process, and as a result, engineers usability, into the product. 752 Mital and Pennathur Figure 3 System-centered approach to design. Copyright © 2000 Marcel Dekker, Inc. 1.3 THE PROCESS OF SOLVING HUMAN± MACHINE INTERFACE PROBLEMS Even though displays and controls are the ®nal means of information exchange between humans and machines in a system, the actual design of the hard- ware and software for displays and controls comes only last in order, in the process of solving human± machine interface problems. The other key steps in this process include user-needs analysis, task analysis, situation analysis, and function allocation decisions, after which the modes of information presentation and control can be decided. In the following sections, we discuss each of these steps. 1.3.1 User-Needs Analysis The goal of user-needs analysis is to collect informa- tion about users and incorporate it into the design process for better design of the human±machine interface. User-needs analysis typically involves the following activities: characterization of the user, characterization of the task the user performs, and characterization of the situation under which the user Human Interfaces for Automated Systems 753 Figure 4 Human-centered approach. Copyright © 2000 Marcel Dekker, Inc. has to perform the task. What follows are guide- lines and methods for performing each of these three activities prior to designing the system. 1.3.1.1 Characterization of the User Table 1 provides a user characterization checklist. Included in this checklist are questions to elicit infor- mation about the users, information about users' jobs, information about users' backgrounds, information about usage constraints, and information about the personal preferences and traits of the users. As is obvious from the nature of the questions in the checklist, the goal of collecting such information is to use the information in designing a usable system. 1.3.1.2 Characterization of the Task Characterization of the tasks users have to perform to attain system goals is done through task analysis. Task analysis is defned as the formal study of what a human operator (or a team of operators) is required to do to achieve a system goal [6]. This study is conducted in terms of the actions and/or the cognitive processes involved in achieving the system goal. Task analysis is a methodology supported by a number of techniques to help the analyst collect information about a system, organize this information, and use this information to make system design decisions. Task analysis is an essential part of system design to ensure ecient and eective integration of the human element into the system by taking into account the limitations and cap- abilities in human performance and behavior. This integration is key to the safe and productive operation of the system. The key questions to ask when performing task ana- lysisactivitiesareshowninTable2.Thetaskanalysis methodology ®nds use at all stages in the life cycle of a systemÐfrom initial conception through the prelimin- ary and detailed design phases, to the prototype and actual product development, to the storage and demo- lition stage. Task analysis is also useful for system evaluation, especially in situations involving system safety issues, and in solving speci®c problems that may arise during the daily operations of a system. Task analysis can be carried out by system designers or by the operations managers who run the system on a day-to-day basis. 754 Mital and Pennathur Table 1 User Characteristics Checklist Data about users What is the target user group? What proportion of users are male and what proportion are female? What is average age/age range of users? What are the cultural characteristics of users? Data about job What is the role of the user (job description)? What are the main activities in the job? What are the main responsibilities of the user? What is the reporting structure for the user? What is the reward structure for the user? What are the user schedules? What is the quality of output from the user? What is the turnover rate of the user? Data about user What is the education/knowledge/experience of the user relevant to the job? background What are the relevant skills possessed by the user? What relevant training have the users undergone? Data about usage Is the current equipment use by users voluntary or mandatory? constrains What are the motivators and demotivators for use? Data about user What is the learning style of the user? personal What is the interaction style of the user? preferences and What is the aesthetic preference of the user? traits What are the personality traits of the user? What are the physical traits of the user? Adapted from Ref. 5. Copyright © 2000 Marcel Dekker, Inc. While many dierent task analysis techniques exist to suit the dierent design requirements in systems, our primary focus here is on techniques that help in design- ing the interface. The key issues involved in designing a human interface with automated equipment are asses- sing what will be needed to do a job (the types of information that human operators will need to under- stand the current system status and requirements; the types of output that human operators will have to make to control the system), and deciding how this willbeprovided.Table3providesasummaryofthe important activities involved in the process of interface design and the corresponding task analysis technique to use in designing this activity. We present brief sum- maries of each of these techniques in the following sections. The reader should refer to Kirwan and Ainsworth [6], or other articles on task analysis, for a detailed discussion of the dierent task analysis tech- niques. Hierarchical Task Analysis. This enables the analyst to describe tasks in terms of operations performed by the human operator to attain speci®c goals, and ``plans'' or ``statements of conditions'' when each of a set of operations has to be carried out to attain an operating goal. Goals are de®ned as ``desired states of Human Interfaces for Automated Systems 755 Table 2 Checklist for Task Analysis Activities Goals What are the important goals and supporting tasks? For every important task: Intrinsics of the task What is the task? What are the inputs and outputs for the task? What is the transformation process (inputs to outputs)? What are the operational procedures? What are the operational patterns? What are the decision points? What problems need solving? What planning is needed? What is the terminology used for task speci®cation? What is the equipment used? Task dependency and What are the dependency relationships between the current task and the other tasks and systems? criticality What are the concurrently occurring eects? What is the criticality of the task? Current user problems What are the current user problems in performing this task? Performance criteria What is the speed? What is the accuracy? What is the quality Task criteria What is the sequence of actions? What is the frequency of actions? What is the importance of actions? What are the functional relationships between actions? What is the availability of functions? What is the ¯exibility of operations? User discretion Can the user control or determine pace? Can the user control or determine priority? Can the user control or determine procedure? Task demands What are the physical demands? What are the perceptual demands? What are the cognitive demands? What are the envirornmental demands? What are the health and safety requirements? Adapted from Ref. 5. Copyright © 2000 Marcel Dekker, Inc. systems under control or supervision'' (e.g., maximum system productivity). Tasks are the elements in the method to obtain the goals in the presence of con- straints (e.g., material availability). Operations are what humans actually do to attain the goals. Thus, hierarchical task analysis is ``the process of critically examining the task factors, i.e., the human operator's resources, constraints and preferencesÐin order to establish how these in¯uence human operations in the attainment of system goals.'' System goals can be described at various levels of detail (or subgoals), and hence the term ``hierarchical.'' The hierarchical task analysis process begins with the statement of overall goal, followed by statements of the subordinate opera- tions, and the plans to achieve the goal. The subordi- nate operations and the plans are then checked for adequacy of redescription (of the goal into subopera- tions and plans). The level of detail necessary to ade- quately describe a goal in terms of its task elements determines the ``stopping rule'' to use when redescrib- ing. A possible stopping rule could be when the prob- ability of inadequate performance multiplied by the costs involved if further redescription is not carried out, is acceptable to the analyst. Activity Sampling. This is another commonly used task analysis method for collecting information about the type and the frequency of activities making up atask.Figure5showsthestepsinvolvedinactivity sampling. Samples of the human operator's behavior at speci- ®ed intervals are collected to determine the proportion of time the operator spends performing the identi®ed activities. Two key factors for the activity sampling method to work include the requirements that the task elements be observable and distinct from one another, and that the sampling keep pace with the performance of the task. Typically, the analyst per- forming activity sampling, classi®es the activities involved, develops a sampling schedule (these two aspects form the core of the design of activity samp- ling), collects and records information about activities, and analyzes the collected activity samples. Activity sampling has its advantages and disadvantages. Objectivity in data recording and collection, ease of administering the technique, and the ability of the technique to reveal task-unrelated activities that need analysis, are some of the advantages of the method. Requirements of a skilled analyst (for proper identi®- cation and description of the task elements), and the inability of the technique to provide for analysis of cognitive activities are the main disadvantages of the technique. Task Decomposition. This is a method used to exactly state the tasks involved in terms of information con- 756 Mital and Pennathur Table 3 Summary of Task Analysis Activities and Methods Involved in Interface Design Activity Task analysis method Gathering task information Hierarchical task analysis representing the activities within the task Activity sampling Stating required information, actions, Work study and feedback Task decomposition Decision/action diagrams Checking adequacy of provisions for Table-top analysis information ¯ows for successful Simulation completion of the task Walk-through/talk-through Operator modi®cations surveys Coding consistency surveys Identifying links between attributes Link analysis (total system check) to ensure system Petri nets success Mock-ups Simulator trials Provide detailed design Person speci®cation recommendations Ergonomics checklists Modi®ed from Ref. 6. Copyright © 2000 Marcel Dekker, Inc. tent, and actions and feedback required of the opera- tor. Once a broad list of activities and the tasks involved have been generated using either hierarchical task analysis or activity sampling, task decomposition can be used to systematically expand on the task descriptions. The various steps involved in task decom- positionarepresentedinFig.6. Decision±Action Diagram. This is one of the most commonlyusedtoolsfordecisionmaking.Figure7 is an example of a decision±action diagram [7]. The decision±action diagram sequentially proceeds through a series of questions (representing decisions) and pos- sible yes/no answers (representing actions that can be taken). The questions are represented as diamonds, and the possible alternatives are labeled on the exit lines from the diamond. A thorough knowledge of the system components, and the possible outcomes of making decisions about system components is essential for constructing complete and representative decision± action diagrams. Table-Top Analysis. As the name implies, this is a technique through which experts knowledgeable about a system discuss speci®c system characteristics. In the context of interface design, this task analysis methodology is used for checking if the information ¯ows identi®ed during the initial task analysis and task description, is adequate for successful task com- pletion. Table-top analysis, hence, typically follows the initial hierarchical or other forms of task analysis which yield task descriptions, and provides informa- tion input for the decomposition of the tasks. A num- ber of group discussion techniques exist in practice, including the Delphi method, the group consensus approach, the nominal group technique, etc., for con- ducting table-top analysis, each with its own merits and demerits. Walk-Through/Talk-Through Analysis. These ana- lyses involve operators and other individuals having operational experience with the system, walking and talking the analyst through observable task com- ponents of a system in real time. Walk-through is normally achieved in a completely operational system or in a simulated setting or even in a mock-up setting. Talk-through can be performed even without a simula- tion of the systemÐthe only requirements are drawing and other system speci®c documentation to enable the analysts to set system and task boundaries while per- forming the talk-through analysis. For more informa- tion on walk-through and talk-through analyses, refer to Meister [8]. Human Interfaces for Automated Systems 757 Figure 5 Activities involved in activity sampling. Copyright © 2000 Marcel Dekker, Inc. Operator Modi®cation Surveys. These surveys are performed to gather input from the actual users, (i.e., the operators) of the system, to check if there will be diculties in using the system, and of what types. This checking of the adequacy of the interface design of the system from the users' perspective is done through surveys conducted on similar already operational systems. In general, operators and other users of systems maintain and provide information on design inadequacies through memory aids, such as their own labels on displays to mark safe limits, per- ceptual cues, such as makeshift pointers, and organi- zational cues, such as grouping instruments through the use of lines. These makeshift modi®cations done by the operators indicate design de®ciencies in the system, and can be planned for and included in the redesign of the existing system or in the design of a new system. Coding Consistency Surveys. These surveys are used to determine if the coding schemes in use in the system are consistent with the associated meanings, and if and where additional coding is needed. The recommendation when performing coding consis- tency surveys is to record the description of the loca- tion of the item, a description of the coding used for that item (intermittent siren sound), a description of any other coding schemes used for that item (inter- 758 Mital and Pennathur Figure 6 The task decomposition process. Copyright © 2000 Marcel Dekker, Inc. [...]... warning of danger) Typography depends on factors such as the stroke width of the alphanumeric character (ratio of thickness of the stroke to height of the character), the width-to-height ratio of the character, and the type style Table 14 also provides accepted guidelines, based on research, for size of characters, case, layout of characters, and for reading ease of alphanumeric characters Some examples of. .. simple paper-and-pencil simulation, to a mock-up of a system that may or may not be dynamic, to a dynamic simulation which will respond in real time Whatever the method of simulation used, the key consideration in simulation studies is the trade-o between the ®delity of simulation (deciding the features of the system that need ®delity is an issue too), and the cost of involved in building high-®delity... value, trend, rate of change, direction of change, or other similar aspects of a changeable variable Information on the status of a system, information on a one of a limited number of conditions, and information on independent conditions of some class Information on emergency or unsafe conditions, information on presence or absence of some conditions Pictorial or graphic representations of objects, areas,... printed safety signs) A number of other types of information are also recognized in the literature Table 13 provides a list of these types along with a brief description of the characteristics of these types of information In the following sections, we discuss recommendations for the design of dierent types of visual and auditory displays (we restrict our attention in this chapter only to these two common... sides of the pedal) 8 (between the outer sides of the pedal) 2 (between the inner sides of the pedal) 6 (between the outer sides of the pedal) 6 (between the inner sides of the pedal) 10 (between the outer sides of the pedal) 4 (between the inner sides of the pedal) 8 (between the outer sides of the pedal) Randomly with one foot Sequentially with one foot Sequentially with one foot Adapted from Ref 89. .. pi is the probability of occurrence of event i The average information (Hav conveyed by a series of events having dierent probabilities is given by Hav pi log2 1=pi where pi is the probability of the event i Just as a bit is the amount of information, redundancy is the amount of reduction in information from the maximum due to the unequal probabilities of occurrence of events Redundancy... representation of an action or an event The literature recommends the use of no more than ®ve geometrical shapes, as using more than ®ve will lead to diculty in discrimination of the dierent shapes [81] While a total of 24 dierent angles of inclination (of characters) are available if coding is to be done by using angles of inclination, the recommended limit is 12 [82] Using this form of coding has... trend in the variable, or a rate of change of the variable Also, qualitative Copyright © 2000 Marcel Dekker, Inc Mital and Pennathur displays can be used to determine the status of a variable in terms of predetermined ranges (whether the fuel tank is empty, full, or half-full), or for maintaining a desirable range of values of a variable (such as speed) The most common forms of presenting qualitative information... static positioning (maintaining a speci®c position of a body member for a speci®ed period of time) In addition, certain theoretical models of human motor responses explain the control aspects of human responses based on only two fundamental types of movementsÐfast and slow Closed-loop theories [ 59, 60], whether the movement be fast or slow, use the concept of sensory feedback (sensory information available... understanding of response time of the human is essential for good design of the tasks involved in human interaction with automated systems Response time is, in general, composed of reaction time, and movement time Reaction time is de®ned as the time from the signal onset calling for a response, to the beginning of the response Simple reaction time (reaction time in the presence of a single source of stimulus) . for a two-way exchange of information. The recent view of the human±machine system, resulting out of advances in computerized informa- tion systems, sees the human operator as a super- visory. limitations. 1.1.3 Scope of This Chapter Even though displays and controls, and their eective design, are fundamental to the ecient and error-free operation of the system, a number of important activ- ities. notes about the scope of this chapter: due to the vast literature available in the area of design of human±machine systems, our emphasis in this chapter is on the breadth of coverage rather than