Taking into Account Human Error when Assessing the Impact of Dependability on Usability
Résumé
Engineering interactive systems in safety-critical contexts, such as digital aircraft cockpits, is a challenge as there is a need to address adequately both dependability and usability. On one hand, the dependability of the interactive system and its related hardware and software components must be consistent with the criticality of the functions to be controlled and their required DAL (Development Assurance Level) levels [10]. On the other hand, the system's usability must be guaranteed so that the user can control the system and accomplish all the tasks she/he has to perform to operate the critical system safely. In a previous paper [6], we proposed a generic approach for assessing the impact of fault-tolerant architecture on dependability and usability. However, this approach considered the user as perfect i.e. the tasks being performed without errors. This paper extends that generic approach by taking into account in a systematic way human errors. We demonstrate that human errors may be identified in a systematic way using knowledge, techniques, and tools from the human factors and human-computer interaction communities. Using four fault-tolerant architectures in the area of interactive cockpits we show how the approach may be used and the benefits which come from using it.