29/11/2013

Human Factors in Multimodal Information Management

In the past twenty years, computers and networks have gained a prominent role in supporting human communication. This constitutes one of the most remarkable departures from their initial role as processors of large amounts of numeric data, in business or science, or as controllers of repetitive industrial operations. However, to offer truly innovative support to human communication, computers had to demonstrate that they could achieve more than what telephone calls or videoconferencing could do.


The design of advanced information management systems does not only represent a technical challenge for computer scientists and engineers, it also represents a difficult task for human factors specialists. The job of human factors specialists is to help design systems such that they are safe, efficient, and comfortable to use (Helander, 2006). In multimodal information management, a particular challenge for human factors is associated with the design of such systems for teams rather than individual users (Bowers et al., 2006). The change from one-person use to multi-person use has a number of important implications since it requires examining additional issues that are not relevant for single-user system usage. This includes the need for communication, shared mental models (i.e., several users need to have the same understanding of the current situation), and allocation of tasks between users (i.e., who is best suited to perform a given task).

Multimodal information management systems represent an enhancement of video conferencing since they also allow for the integration of task representations into the overall environment. This task representation adds an important element to the technical system, which has not yet been addressed by the present body of research concerned with video conferencing. The task representation emphasizes the importance of the user directly interacting with technology as an integral part of task completion (e.g., entering system commands or taking display readings) rather than simply using it as a device for simply mediating communication between humans as it is the case for video conferencing where the interaction is limited to verbal and non-verbal communication.

Prominent research topics in human factors
Human factors research covers a wide range of subject matters that are addressed in a wide range of application domains. Handbooks of human factors research provide testimony on the high level of research activity in the field (e.g., Salvendy, 2012, Helander, 2006). This includes coverage of many subject matters such as workplace design, illumination, affective engineering and automation, but also coverage of many application domains, such as aviation, process control, car navigation, or consumer product design. Two examples of each, namely automation as a subject matter and consumer products as an application domain, are considered in more detail below to demonstrate what kind of research is typical in each area and how human factors can contribute to the advancement of knowledge.

Automation
The rapid technological progress in computing science and engineering over recent years has resulted in increased automation of work systems but also of systems, used in the domestic domain (e.g., autopilot in aircraft, car park assistant, automatic window shutters). This progress has led to a number of problems for the human operator when managing such systems. A primary problem in high-level automation is the ensuing decrement of operator skill due to the lack of practice of critical competences, because the machine has taken over many tasks. A too high level of opaqueness in system operation is also a typical observation in highly automated systems, which poses problems when automatic systems fail and operators suddenly have to adopt manual control without being well informed about the current state of the system (“out-of- the-loop” problem).

These two examples demonstrate the implications that automation may have for human operators. Related to this is a major problem, which concerns the question of what level of automation should be chosen to match best the needs of the operator (e.g., full manual control by human, fully automated control, or intermediate level at which the human can veto a decision taken by the machine). Modern approaches to automation, that is, adaptable and adaptive automation, allow for changes to be made to the level of automation if the need arises (e.g., if the human operator suffers from excessive workload levels, the level of automation will increase). The important question is whether it should be the human or the machine that decides if any such change in level of automation should be initiated. There is evidence that the human may not be the best judge of adapting levels of automation (Kaber and Riley, 1999) because they tend to rely more strongly on manual control than actually needed (Sauer et al., 2013). Conversely, the algorithm that initiates changes in level of automation (if the ultimate authority is afforded to the machine) may be highly sophisticated and difficult to develop in practice (Sauer et al., 2012).

Extensive and rigorous human factors testing is needed to provide answers to these questions and to develop solutions that can be successfully implemented in real systems. The research issues surrounding automation have increasingly found their way into the design of technical systems that are used
outside work contexts.

Consumer product design
The design of products to be used in the domestic domain or for leisure purposes is faced with a number of challenges that are different from those faced in work system design. While domestic products generally enjoy a lower level of complexity, users are less well prepared for operating devices than in work contexts. This problem is due to a number of factors in which the domestic domain differs from the work domain (Sauer and Ru ̈ttinger, 2007). First, in the domestic domain, users are not selected on the basis of competence, following a formal selection procedure. Second, users do not receive any formal training to be allowed to operate consumer products. Third, in the domestic domain tasks are typically self-defined by the user and not prescribed by domain experts. Fourth, user performance in the domestic domain is usually not subject to supervision by subject matter experts, resulting in performance feedback. Overall, these factors result in a considerable degree of diversity in user behavior, which by and large can only be modified by product design since other means of influencing user behavior (i.e., selection, training, task design, feedback) are hardly available in the domestic domain.

The main lesson to be learnt is that product designers have to focus more strongly on the actual design of the device. However, the way products are used and evaluated is not only affected by product functionalities. ingly, the aesthetic properties of a product do not only affect attractiveness ratings (which one would expect) but also user behavior and ratings of product usability.

A series of studies showed that usability ratings are increased when the product is considered to be aesthetically pleasing (e.g., Sauer and Sonderegger, 2009, Hartmann et al., 2007, Sonderegger and Sauer, 2010). Further studies demonstrated that aesthetic design qualities may have an effect on performance (e.g., Ben-Bassat et al., 2006, Moshagen et al., 2009, Sonderegger and Sauer, 2010), though the direction of the effect differs across studies.

To learn more

Extract of the book: Interactive Multimodal Information Management
Under the direction of Hervé Bourlard and Andrei Popescu-Belis
Published by the Presses Polytechniques et universitaires romandes

Les commentaires sont fermés.