Using human-compatible reference values in design of cooperative dynamic human-machine systems

Abstract

This is a conceptual paper about design of cognitive systems. The design of cooperative dynamic human-machine systems, such as driver assistance systems within a vehicle being part of the traffic, is a challenging task. The result has to be easily comprehensible for the user, although the overall system complexity might be rather high. Firstly, the driver assistance systems are able to perform a complex but partially hidden behavior. An Adaptive Cruise Control system (ACC), for example, can perceive relevant aspects of the current traffic situation and can decide autonomously about the longitudinal maneuver, such as approaching or maintaining speed. In parallel, it performs singular actions, such as deceleration to avoid collisions. However, the matching of technical and design constraints of the machine behavior and natural human behavior can be difficult, particularly in time- and safety-critical situations, for instance during emergency braking or evading. And secondly, the overall system complexity increases caused by the high number of different assistance systems already on the market. In some cases they are not well integrated and the driver has to interact with each of them separately. To design well usable cooperative dynamic human-machine systems and interaction, a usability improving and system integrating approach is required. In this paper, we describe such an approach based on the use of human-compatible reference values. The main reference value that we use in the design of driver assistance systems is named ‘action tension’. We show exemplary how such a value can be derived from well-known and accepted scientific concepts, how it can be operationalized and evaluated within a usability assessment and how it can be used in the design of an integrated driver assistance system in case of an approach, brake and evade driving scenario.