FMHAI · Formal Analysis and Modeling of Human-Automation Interaction

Formal Analysis and Modeling of Human-Automation Interaction

The initial focus of this project was to develop a formal language for the specification, design, and evaluation of automotive systems and their user interfaces. The project began by identifying the needs and requirements of user interaction designers and those of GM. This phase of the project involved extensive discussions with designers and human factors professionals who were actually working on the design of next generation infotainment systems at GM. After about a half a year of mutual work including trips to Detroit, a specific automotive language was developed. This language is based on David Harel’s Statecharts language. It was tested on several new designs and evolved to become a specification language used by GM designers to augment their text-based specifications. A joint paper with a GM designer on how to develop formal specifications for human-machine interaction was delivered at an automotive trade conference in Detroit (Degani & Gellatly, 2011). Another conference proceeding paper described the use of the language to model and identify opportunities in the design of automated driving aids such as a lane centering system and advance cruise control system (Heymann & Degani, 2013). A book chapter detailing the general approach for modeling human-machine interaction was published in a handbook of cognitive engineering (Degani & Heymann, 2013). The second objective was to use the language to identify design deficiencies in automotive human machine systems. This objective was the outcome of using the formal language for specification and attempting to highlight the “shape and format” of both design mistakes and design opportunities. The effort resulted in a set of design properties and also several heuristic design principles to aid engineers and designers in developing infotainment systems. A joint paper with a GM lead designer and the university collaborator documenting this effort and was presented at the 2011 IEEE Systems, Man, and Cybernetics conference (Degani, Heymann & Gellatly, 2011). The general approach and modeling methodology was also presented at an academic conference as an example of an industry contribution (Jamieson et al., 2012). The third objective concerned the identification of “good” design properties for human-machine interactions. To explore the utility of this approach, a research project was conducted to evaluate interfaces for automotive displays that have some elements of automated driving in them. The study measured and quantified the dimensions of “good” integrative display design. The results were presented at the last annual meeting of the Human Factors and Ergonomics Society (Shmueli et al., 2013). This approach to identification of good design properties for the organization of information is based on a theoretical construct that involves a pyramid- like structure of abstraction, integration, and configuration of information. Objectives number four concerns the development of design tools to support designers at GM. An initial effort was undertaken to develop a graph-based methods to quantify the interrelationships between informational elements on the main display of the car. The theoretical foundation for the tool are described in Shmueli et al., 2013, and the actual tool is in used by GM designers.

Data: CORDIS, © European Union