The main topic of this paper is a methodology of designing driver support systems. A driver support system (co-pilot) is required to warn the driver about immediate dangers, such as unexpected obstacles on the road, to provide him with help when he is performing complex manoeuvres, such as overtaking or passing an intersection, and maybe to intervene in some extreme cases, when there is no other possibility of avoiding an accident. Knowledge about the dynamically changing road situations has to be continuously extracted from sensor data. The process of forming high-level, symbolic descriptions of traffic situations in real-time, called here characterization, is one of the crucial elements of the analysis and evaluation of the current traffic scenario. The paper focuses on some chosen aspects of driver support system design methodology, namely on: 1) a solution for the real-time characterization problem , 2) a choice of knowledge representation formalisms required by various co-pilot functions; and 3) adoption of systematic software engineering principles, and especially of layered software architecture and software engines facilitating the design and prototyping phases. (A)
Abstract