System-Ergonomic Design of Cognitive Automation

Why this book? Simply because it is due. Cognitive automation and its syste-mergonomic introduction into work systems have been advanced in the meantime to such a degree that already applications for operational work systems are slowly becoming reality. This interdisciplinary book is meant for designers of work systems and associated machines who are interested in this modern approach and its implementation.

If we talk about vehicle guidance and control to people who have their own experience in being often directly involved in such a process like you and us as car drivers or like airplane pilots, they may at first think about what have been the challenges and problems when driving a car or flying an aircraft, and how they mastered the demanding situations. If we talk about vehicle guidance and control with control engineers, they may look at the same thing from a different perspective. At first they may think about technical systems which are or can be installed in the vehicle in order to support the vehicle operator, usually to make his job an easier one. Either view does not cover the whole issue, but both are important ones on their own right. Before we put both these views together into a single comprehensive one, this chapter will deal just with the engineer’s view with the focus on operational systems in aircraft and automotive vehicles.

In order to initiate new developments of vehicle guidance and control systems which go beyond the state of the art, it appears to be worthwhile to put in front some basic facts and considerations. In particular, it makes sense to clarify how a system is structured which enables the process of vehicle guidance and control and what we can learn from human cognition in order to make use of it for the enhancement of vehicle guidance and control. In essence, this leads to an understanding of two general concepts, that of work and pertinent work systems, and that of cognition. These two concepts compose the basis for the considerations of system design which will be outlined thereafter. This chapter therefore will start with a discourse on the concept of work and its implications, if we consider the process of vehicle guidance and control as a work process. Subsequently, the main features of (human) cognition are outlined in order to capture its design potentials, if one makes use of introducing artificial cognition in the work process.

Considering the system design in work environments, engineers as opposed to psychologists are interested to build adequate tools and support systems for human operators for the sake of good, satisfying work results. The engineers’ focus lies on the issue of system synthesis in the work environment rather than system analysis, thereby working around the human operator, who is considered as a given system element. As to the human component, work system designers can only make a good job, if they do not ignore what was already found about human factors, including human cognition. These findings are indeed of great value for the work system design, i.e. to define when and how support for the human operator by artificial means, in particular automation, is to be considered for the sake of both work process performance and that there is a good feeling on the human’s side that he has everything well under control.

The preceding chapters have established a kind of foundation for a systematic approach to achieve a system-ergonomic design of work systems. It was pointed out that there are two levels of automation which can be made use of, conventional and cognitive automation. As to cognitive automation, the artificial cognitive unit (ACU) was introduced as a particular kind of agents, based on a conception of the artificial cognitive process in some analogy to human cognition. Again, there are two possible modes of ACUs to account for in a work system design (dual-mode design), supporting cognitive units (SCUs) as part of the operation-supporting means in the work system, and operating cognitive units (OCUs) as members of a team making up the operating force of the work system.

This chapter shall contribute to illustrate that there are already implementations of either type of ACUs, SCUs and OCUs, as well as their integration in work systems of vehicle guidance and control. In the following sections a number of them will be described in some detail, starting with SCUs.

Referring to the functional framework in Figure 30 for human cognition shown in Chapter 3.2.4 and the process of artificial cognition in an ACU described in Chapter 4.2.1 some implementation examples of crucial functional components of cognitive automation will be presented in more detail in the following. The choice of these implementation examples is made in a way to

The order of presenting the examples of component implementations, which will be presented in the following, is essentially chosen from a bottom-up perspective. About all of them are covering more than one cell of the array shown in the functional framework of Figure 30. That means that lower-level functions as enablers and behaviourally lower levels come prior to higher-level functions and higher-level behaviour. It turns out that there is at least one implementation example associated to each cell of functional and behavioural levels concerned. Furthermore, it should be kept in mind that, in the first place, all high-level functions from skill-based through concept-based behaviour, which are essentially identical to those of ACUs, are determined by the a-priori knowledge to be represented and implemented. Therefore, the knowledge management is the main issue of the following implementation examples.

The operationalisation of cognitive automation in work systems is probably the most controversial topic of this book. This is because there are hardly any automated functions of that kind in operational use today. Because of the lack of engineering experience there is great hesitation in most industrial design teams, in particular as to the software certification of cognitive automation, since this software seems to them even more complex than that known for conventional automation. This concern, however, is definitely not justified. The strongly praised reliability of human individuals when acting is mainly founded on the fact that their motivational contexts and their knowledge about the work objective provide the criteria needed to appropriately carry out and monitor his activities. The same is valid for artificial cognition in work systems, if it works on the same basis of motivational contexts and knowledge about the work objective. In order to illustrate this point, an example for an implementation of cognitive automation is given in Chapter 7.2.1 which is dealing with self-monitoring on that basis. Another issue is the lack of development standards. This is still a rather open field. There is not much application literature around. Therefore, at first a development framework, called COSA (COgnitive System Architecture), will be described in the following to give some guidance for those who otherwise would hesitate to take care of that and to make the start on their own.