Search

This article presents an overview of COSA, a cognitive system architecture, which is a generic framework proposing a unified architecture for cognitive systems. Conventional automation and similar systems lack the ability of cooperation and cognition, leading to serious deficiencies when acting in complex environments, especially in the context of human-computer interaction. Cognitive systems based on cognitive automation can overcome these deficiencies. Designing such artificial cognitive systems can be considered a very complex software development process. Although a number of developments of artificial cognitive systems have already demonstrated great functional potentials in field tests, the engineering approach of this kind of software is still a candidate for further improvement. Therefore, wide-spread application of cognitive systems has not been achieved yet. This article presents a new engineering approach for cognitive systems, implemented by the COSA framework, which may be a crucial step forward to achieve a wide-spread application of cognitive systems. The approach is based on a new concept of generating cognitive behaviour, the cognitive process (CP). The CP can be regarded as a model of the human information processing loop whose behaviour is solely driven by "a-priori knowledge". The main features of COSA are the implementation of the CP as its kernel and the separation of architecture from application leading to reduced development time and increased knowledge reuse. Additionally, separating the knowledge modelling process from behaviour generation enables the knowledge designer to use the knowledge representation that is best suited to his modelling problem. A first application based on COSA implements an autonomous unmanned air vehicle accomplishing a military reconnaissance mission. Some of the application experiences with the new approach are presented. [ABSTRACT FROM AUTHOR]

Technological progress keeps on changing our lives and our work. We can only improve this mutual influence between human and technology if we understand the impact of technology on the work process, and if we are also able to share this understanding with other people in the design process. Quantitative assessment methods like workload offer at least a keyhole to the work process, but only additional concepts like situation awareness or usability in an appropriate combination with qualitative aspects can open this keyhole into a wider window. The pointillist approach uses a powerful metaphor, borrowed from the arts, to combine quantitative with qualitative aspects systematically. Based on this approach, a tool for human factors analysis was built and used to analyse the interaction between a pilot and an assistant system in a simulator. Three examples illustrate the possible interplay between quantitative and qualitative aspects of the approach. [ABSTRACT FROM AUTHOR]