Management of Man and Machine
Management of Man and Machine
Human intervention of the production processes has undergone a big change after automatic and computerized controls have been introduced for the production processes. A large number of activities previously done by human beings have been taken over by the automation. But this has not eliminated the need for operator for the running of the equipment/process though his role has changed a lot with the automation of the process.
Today mass production would not exist without the usage of automated and flexible manufacturing processes. These automated processes need machines and equipments which require human intervention for controlling them. Close and harmonious interaction by operators with their machines is a necessity for the productive output. An integrated and coordinated communication between machines and the men operating them is needed for the productive output.
The complexity of industrial processes has greatly increased during the last few decades. This tendency has originated due to a number of reasons, such as (i) the enlargement of the scale of the modern plants, (ii) the required specifications dealing with the product quality, (iii) the need for the energy conservation, (iv) the requirements for the environmental pollution control, (v) the necessity of safety in the plant, and (vi) the progress in process control and informatics creating totally new possibilities.
This essential change in the process operation has led to the definition of new human operator tasks. In the last thirty years, human manual control has become much less important and human supervisory control has been developed as the main concept for man and machine interactions. The tasks of the human supervisor are now predominantly cognitive ones, and contain at least the following six subtasks namely (i) the monitoring of all data presented to the human supervisor, (ii) the learning and interpretation of the data presented, (iii) the process tuning or set-point control or teaching of the process in the normal circumstances, (iv) the intervention into the process for instance during abnormal process conditions, (v) the fault management during malfunctioning of the plant, and (vi) the planning such as for starting-up and shutting down the plant.
All the above changes in complexity and requirements of industrial process plants, in automation concepts and technologies related to new systems, computer and software engineering approaches, as well as in the tasks of the human operator have shown the evidence of a number of theoretical problems in man-machine systems more clearly. Concepts of cognitive engineering and human-centred design approaches have evolved as possible answers to these problems. These approaches suggest that human information processing behaviour as well as knowledge and goals structures of the human operator are required to be investigated. The results need to be applied in advanced automation and decision support systems as well as in advanced man-machine interfaces, in order to guarantee enough flexibility and job satisfaction for the human operator which are prerequisites for the safe systems operation.
Improvements in man-machine systems are indispensable because of the dealing with the developments which are trying to contribute to better man-machine systems in the industrial plants.
Methodologies to evaluate and to validate human behaviour have been developed in great detail in manual control. However, the state of the art is totally different in case of the supervisory control. The high complexity of the plant and the vague definition of the subtasks of the human supervisor are of crucial significance. It is therefore necessary that some special attention is focused on the phenomenon of complexity before the evaluation and validation of supervisory control behaviour.
Complexity is directly related to the combination of four factors namely (i) large numbers, (ii) diversity, (iii) coupling, and (iv) interaction. The factor of large numbers deals with the number of functions an automation system satisfies, whereas the factor of diversity may be interpreted in two different ways namely (i) it either means the different functions, or (ii) it means flexibility. The factors of coupling and interaction are synonymous with static and dynamic interactions. Hence, complexity is mainly dependent on two factors namely (i) the number of functions and the interaction, and (ii) the design of a system with the optimization of a criterion where complexity and flexibility are weighted.
The problems arising when an operator has to take over a complex process during a monitoring task (the vigilance problem) have to be clearly described. To take-over of the process control functions is especially problematic when the system runs into an unknown state. Training of operator in a simulator is one possible consequence, better is permanent on-line control in the real process.
High skilled operators tend to lose the potential to be aware of the whole process in case of automated controls. They need a special qualification to get open minded. If incongruity is too low, then humans try to increase the contextual complexity. This perspective allows having an alternative interpretation of human ‘failures’ in inescapable situations with information under-load (e.g., process monitoring in a steady-state). To increase the signal rate of the machine system artificially is not an appropriate design strategy for man-machine systems. Job rotation and job enrichment can help to reduce information under-load, but not for a long time. Depending on the learning rate of the operator, it is also necessary to be aware of the monotony problem. The best solution is to involve the operator in the task solving process, especially when the task is a ‘complete task’. Operators need to have on-line control over the real process. To satisfy the human need for variety (and optimal information) the work system must be flexible and individualistic. Of course, this demand leads to difficulties in complex system design.
Man machine management attempts to rationalize relevant attributes and categories that emerge from the use of machines. Four main principles, that is, safety, performance, comfort and esthetics, drive this rationalization along with four human factors which are physical (that is, physiological and bio-mechanical), cognitive, social or emotional (Fig 1).
Fig 1 attributes and categories emerging from the usage of machine
Current machines heavily rely on the cognitive skills of their users, who acquire and process data, make decisions and act in real-time. Machines are becoming more complex even if the goal of the designers is to facilitate their use during normal operations since problems do happen in routine as well as abnormal conditions. This is why human reliability needs to be taken into considerations carefully from two points of view namely (i) humans have limitations, and (ii) humans are unique problem-solvers in unanticipated situations. There are three levels of human behaviour which are (i) skill, (ii) rule, and (iii) knowledge.
Today, human operators mostly work at knowledge-based level where interpretation has become an important work process. Basic operations are delegated to the machine, and humans progressively become managers of (networked) cognitive systems. Humans need to identify a situation when there is no pattern matching (situation recognition) at the rule-based level, to decide according to specified (or sometimes unspecified) goals, and to plan a series of tasks. These are typical strategic activities. Some operators are good at strategic activities, while others prefer to execute what they are told to do. In any case, the control of cognitive systems requires strategic training.
Another view is based on the three-level concept of human cognitive behaviour, where a distinction is made between a target-oriented skill- based-behaviour (SBB), a procedure oriented rule-based-behaviour (RBB), and a goal controlled knowledge-based-behaviour (KBB). This qualitative method is often used in order to classily human operator tasks. It is widely accepted that manual control tasks and intervention tasks in stationary process conditions mainly lead to SBB, whereas monitoring, interpreting, and teaching a plant in stationary as well as in non- stationary conditions are most often RBB. Fault management and planning are not easily to be classified. They require not only knowledge of the tasks to be performed but also appeal to the creativity and intelligence of the human operator. Hence, they lead to KBB. Tab 1 show that different subtasks are performed at different cognitive levels and, as a consequence, simple methods as validated in manual control certainly cannot be expected to be developed for supervisory control.
|Tab 1 Influence of human characteristic on information systems|
|Human characteristic||Constraint put upon automation systems|
|Difficult work with high number of detailed data||Filter data, represent aggregated data and charts|
|Low communication bandwidth, limiting the transfer of information between automation system and man||Allow for the most optimal user-interface, using new technology for input and output of data|
|Hard context switch when going from one task to another||Not demand short, diverse tasks of the human workers|
|Motivation problem when autonomy is too much restricted||Allow human workers for some decision freedom|
|Making mistakes, forgetting input the data||Check for data consistency|
|Slow response time to events||If possible, getting early attention of operator|
|Low predictability of operation times of manual tasks||Take it into account, if possible based upon historical information|
Another aspect that complicates human performance is the often extremely vague definition of tasks, a situation where the human’s creativity is explicitly required. This yields some very contradictory and intriguing situations, i.e. if human creativity can be predicted one can no longer speak about creativity in fact one might say that the cognitive level of KBB is shifted to that of RBB.
Normally a good and practical automation system design for machines is to be robust to errors. However errors are important from the point of view of stimulating human creativity. A design is always the result of some kind of optimization according to a more or less well-defined criterion, generally a criterion where performance of human and automation system are weighted with the costs, i. e. mental work load. In addition, the question to what extent the internal representation i.e. the knowledge available to the operator, plays a role in mental workload and also in performance is quite important. The whole issue is of high practical use for evaluating alternative man-machine system designs, particularly also those with computer-aided decision support.
The characteristics of manufacturing, decides that it has to be considered as a discrete production. The number and interrelations of control parameters and constrains decide that discrete production can be treated as a complex system. The main features of such systems are as follows.
- Large number of components
- Multiplicity of types of components
- Highly coupled components
- Presence of disturbances
Full automation of control functions in production appears to be impossible without losing flexibility and malfunctions’ resistance, what is the consequence of the complexity of present manufacturing systems. High performance, products quality, and systems flexibility required from one side and complex problems solving, and the experience based learning from the other side require strict co-operation of the information technology and human factor. However, to be effective such co-operation has to take into account both human and information systems features.
The human and information system co-operation is strongly dependent on behaviour of a man and machine automation system. Human operator has very unique natural abilities that decide about his important role in manufacturing. The most important roles are namely (i) experience based self-learning, (ii) adaptation to new situations, (iii) high abilities of manual manipulation, (iv) very efficient sense feedback (eyesight, hearing, etc.), (v) possibility of innovative solution application, (vi) foreseeing of controlled system behaviour based on current observations, and (vii) possibility of reactions on unforeseen situations. These features decide about very strong advantages of man in both machine control and management level.
The most important drawbacks of human beings are namely (i) limited number of information processed at the same time, (ii) difficulty or impossibility of quickly changing process control, (iii) no deterministic behaviour, and (iv) growing malfunctions if tasks are changing too often or are monotonous. The human efficiency decreases when the aim of work is not understood or the operator cannot take decisions about it.
Some of pointed drawbacks can be eliminated or their influence on manufacturing process can be limited by proper work organization. Operator is to know the aim of his work and is to be able to take even limited decisions concerning his work place. In the case of human-machine automation system co-operation the graphical interface need to be designed with taking into account its usefulness and easy operation by particular operator or groups of operators.
However, computer applications and graphical interfaces very often are designed by programmers that do not know specific of particular workplace and human- machine automation system co-operation issues.
Present machine automation information systems have very high speed and processor capacity, high accuracy, and ability to store and process large amount of data. The main drawbacks of these information systems are namely (i) work according to stiff algorithms, (ii) lack of innovative reactions, and (iii) real intelligence defined as skillfulness of new algorithms creation. Important limitation in the case of human co-operation is a necessity of precise task definition and necessity of accurate data input. Human- machine system co-operation is based on the integration of two systems with completely different features. The intelligent man being able to take intuitive, actions, solve diverse problems and learn by experience is on one side. On the other side is the machine automation system that can process large amount of data in short time, has detailed information and short response time. Such integration can be risky, but it gives possibility of complementary co-operation in the case of proper system designing.
Nowadays most of machines used in manufacturing have special control systems. Such a system can control realization of most of machine functions. The control systems can be uncomplicated, based on simple PLC (Programmable Logical Control) or based on advanced multilevel hierarchical controls. Such systems can be responsible for control and supervision of the advanced machines and processes. However, all kinds of machine control systems have to be supervised by human operator. He is necessary in complex problems solving, like in the case on malfunctions and break downs.
Machine operator may co-operate with machine or process through the system control and outside the control system, as well. There are six kinds of interactions between the human operator and computerized automation system controlled machine.
- Operator orders the process indirectly through the automation system
- Operator receives the process status indirectly
- Operator asks for or receives information from the automation system
- Process asks for or receives information from the automation system
- Operator intervenes directly into the process
- Operator observes the process by his own sense
In the above interactions, the first four interactions are controlled by the machine automation system. In other words, the system designer has foreseen them and they have been appropriated in the controlled system and operators requirements. The practically impossible for the control system is the integration of direct operator interactions into the control system of the machine (last two interactions). Such interactions usually occur in the case of the machine or process incorrect behaviour, when the man intervention is necessary.
Man-machine interaction can be realized in different ways, also outside of the control systems. Properly designed machine control systems allow for interactions in predictable situations. They are robust for unpredictable situations like direct human intervention into the controlled machine or process as well. They are especially important in automated machines and monitored machining processes.
The behaviour of human operator
Designers of automation system normally focus their attention on proper material and information processing, but do not appreciate enough influence of human behaviour on the system performance. As a result of such approach the system does not work properly very often, because of man-machine co-operation problems.
Elaboration of thehuman behaviour in manufacturing system is very important. Development of a simulation software as a support for designers of manufacturing systems or its components is possible. However, the problem of incorporation of human behaviour is very difficult. Human behaviour cannot be determined. As a result of it, equations describing human behaviour cannot be incorporated in the system. Hence generally factors characterizing human control work in system are introduced at first. However they are to be minimized. Some of these human behavioral factors are given below.
- Control delay – It can be characterized as a time between abnormal situation appearance and getting system back to the required state.
- Human reaction time – It describes operator’s behaviour, can be characterized as a time between abnormal situation appearance and a control decision taken by the operator.
- System inertia – It is the time between control decision with control action and reaching the proper state by the system
Description of human behaviour requires definition of significant number of non-dimensional factors that point the most important features of human being in manufacturing system. These are (i) human decision correctness, and (ii) problem difficulty.
The interaction process between operators and machines or information systems in manufacturing can be realized by different types of media. Usually, the communication takes place in two directions namely (i) from operator to machine, and (ii) from machine to operator. From the machine or system point of view the interaction from a man to machine is an input, the opposite direction from machine to a man is an output.
The most common communication approach is using of the conventional input/output devices. The input is usually based on the control of human body movements. The most common is the control of fingers and hands movements and are physical movements. However, it is also possible to monitor movements of other parts of body. The input to the system can be based also on the voice commands.
The system output is usually realized by the vision as a transmission medium, but using the audio systems is also popular. In the special machines other mediums, like force or temperature signals and physical feedback can be used, as well.
Communication media suited to needs of the particular process generally does not engage the operator in the information exchange too much. The operator is normally to focus on solving problems appearing in the controlled system, but not on communication with it.
Computer terminals are the most popular input and output devices for advanced man machine interactions. Usually they include: the conventional keyboard, mouse, track-ball, and computer pens.
The most popular output devices for control systems are visual displays. They can be used for presentation of data and images. The data and images from the real and virtual world can be presented on the visual displays. The real world data and images are usually presented on the screens of machine control systems and supervisory manufacturing management systems. Usually, at the same time a few windows presenting different data can be open on the one screen. However the data is to be presented in the simple and easy to understand way. Presentation of images can allow for easier data understanding. It allows also for tele-monitoring, which is very useful in the case of automated manufacturing systems or distributed production.
Presentation of virtual data and images that are results of various simulations can be very helpful for operators. Such solution can be applied for catching the correctness of the manufacturing process control programs. The visual presentation of the simulation results is used very often in the case of whole manufacturing systems simulation for optimizing the system load and production schedule
Various audio displays are used very often in the machine-operator communication. They can present the real sounds, virtual sounds generated by the automation system and data presented in easy to understand, sound form.
The operator-machine communication can be supported by voice commands systems, as well. Such systems can recognize the human speech and sometimes are capable of gaining additional information from the human voice. Very often they allow for the operation time reduction by quicker communication than in the case of the traditional display communication devices. They can be used in dialog management systems, virtual reality systems, but its application on the shop floor can sometimes be difficult because of various kinds of noise.
Other communication media that could be implemented in human-machine information exchange are: force (communication from operator to system), force feedback (communication from system to operator), tactile, temperature, temperature feedback. Force, as a communication way can be used for teaching. However these other pointed media can be applied in manufacturing only in very special applications.