Advances in Human Performance and Cognitive Engineering Research: Volume 2

When designing a joint system for a complex, dynamic, open environment, where the consequences of poor performance by the joint system are potentially grave, the need to shape the machine agents into team players is critical. Traditionally, the assumption has been that if a joint system fails to perform adequately, the cause can be traced to so-called “human error.” However, if one digs a little deeper, they find that the only reason many of these joint systems perform adequately at all is because of the resourcefulness and adaptability that the human agents display in the face of uncommunicative and uncooperative machine agents. The ability of a joint system to perform effectively in the face of difficult problems depends intimately on the ability of the human and machine agents to coordinate and capitalize upon the unique abilities and information to which each agent has access.For automated agents to become team players, there are two fundamental characteristics which need to be designed in from the beginning: observability and directability. In other words, users need to be able to see what the automated agents are doing and what they will do next relative to the state of the process, and users need to be able to re-direct machine activities fluently in instances where they recognize a need to intervene. These two basic capabilities are the keys to fostering a cooperative relationship between the human and machine agents in any joint system.

This chapter focuses upon operators' automation usage decisions (AUDs), choices in which people have the option of relying on automation or employing a manual or less technologically advanced means of control. Misuse, the over-utilization of automation, and disuse, the under-utilization of automation, result from inappropriate AUDs. Three causes of misuse and disuse were identified. Operators may: (1) not recognize that both automated and manual alternatives are available (recognition errors), (2) inaccurately estimate the utilities of the automated and/or manual options (appraisal errors), or (3) knowingly select the alternative with the lowest expected outcome (action errors). A decision-making model was used to organize the literature, identify new areas of inquiry, and examine the effects of a number of variables known to affect AUDs. These included machine reliability and consistency, performance feedback, trust in automation, self-confidence in manual control, perceived utility, and the automation bias.

Aircraft accidents involving airplanes of the three major airplane manufacturers in the database of the National Transportation Safety Board (NTSB) on the World Wide Web were examined for causes. Regression analysis showed that the contribution of automation problems to the total accidents within the 16-year period examined was not statistically significant. However, the contributions of mechanical problems to the overall accident profile was statistically significant F(1,14) = 5.02, p = 0.041844. Some implications of the findings of the study are discussed.

U.S. aviation authorities are particularly curious about how the Gulf Air pilots handled the A320's complex computer-operated control and navigation systems. The planes' software is designed to prevent even the most incompetent pilot from accidentally launching the plane into a fatal stall or dive. But records indicate that some Airbus crashes occurred when pilots misjudged the planes' limitations or made errors entering data into the A320's computer system. And because the system is so complicated, U.S. experts say, if something goes wrong only a mathematical genius could figure out the problem (Mark Hosenball, Newsweek, September 4, 2000).

The purpose of this chapter is to provide a first step towards determining the piloting skills needed for the operation of advanced automated aircraft. Using an information-processing (IP) approach, categorization of the problems observed in interactions between human operators and advanced automated systems were determined. Based on these issues, potential training areas were supplied which may aid to ameliorate many of these interactions between humans and automated systems. Following the identification of potential training areas were brief discussions of the potential efficacy of traditional approaches to performance improvements in these areas. Using the IP framework, numerous problems were identified through examining current literature and research findings, technical reports, as well as incident and accident reports and databases. Five problem areas were identified which, not surprisingly, paralleled the major dimensions of the IP model. These five problem areas included mismatches between hardware and software and the capabilities of the human sensory/perceptual system, increased requirements for vigilance, attention-sharing, and distribution, the development of accurate knowledge in long-term memory, overload of large resource requirements in working memory, and problems relating to decision making. The breakdown of these automation-related problems is important as this step must be completed to determine the foci of any training programs to address these issues. While some believe automation has introduced a whole new array of interaction problems, analyses such as this supply a different snapshot, in our case by using a well investigated information processing model to categorize such problems at their most core level.Investigations should be conducted to determine the effectiveness of the training approaches we recommended. While Wickens' IP model is closed-loop in nature, the model does describe a certain path which information follows throughout the process. Future studies should therefore also examine how training solutions targeting at the earlier stages (i.e. sensation and perceptual) of the model might have a cascading effect in relation to the elimination of problems further along in the processing model.

As we have tried to understand the regional airlines and their current use of automation, it has become clear that there are enough differences between the regional airlines and the transport airlines that research focusing specifically on their needs is required. We have also discovered how difficult it is to define the needs for addressing automation-related topics without better understanding our definitions of automation and how it affects pilot performance. We developed an expanded pilot general task list and definitions of the three categories of automation to help us accomplish our research goals and found that their use results a better methodology for us to explore automation effects and translate them to the needs of the regional airlines.