Ziel der Studie, die im Auftrag der Bundesanstalt fuer Strassenwesen (BASt) durchgefuehrt und zusaetzlich vom Land Niedersachsen gefoerdert wurde, ist es, ueber In-Depth Analysen von Unfalldaten der Braunschweiger Polizei aus dem Jahr 2002 Anforderungen an Fahrerassistenzsysteme (FAS) abzuleiten und den Sicherheitsgewinn abzuschaetzen, den die Einfuehrung derartiger Systeme mit sich bringen wuerde. Bei den In-Depth Analysen wurden aufgrund der Unfallprotokolle die dem Unfall vorausgehenden Fehlhandlungen und ihre Ursachen analysiert. Die Analyse der Ursachen der Fehlhandlung erfolgte in Anlehnung eines Informationsverarbeitungsmodell des menschlichen Handelns und ergibt Hinweise auf die Eingriffsstrategie des FAS. Datenbasis sind 4.258 Unfallprotokolle aus Braunschweig aus dem Jahr 2002 und 185.004 Unfaelle aus Deutschland, die als 50 Prozent-Stichprobe der amtlichen Unfallstatistik des Jahres 2002 vom Statistischen Bundesamt zur Verfuegung gestellt wurde. Bei den In-Depth Analysen wurden fuer 6 Unfalltypen und fuer jeden davon fuer die haeufigsten Untertypen die Protokolle im Hinblick auf Fehlhandlung und Ursache analysiert und gruppiert. Die Analyse zeigt drei grosse Gruppen von Unfaellen, aus denen sich drei Arten von Unterstuetzungsbedarf ableiten lassen. Dabei handelt es sich um Einbiegen/Kreuzen Unfaelle, Fahrunfaelle und Auffahrunfaelle, die entsprechende eigene Assistenzen erfordern. Insgesamt ergibt sich aus den Analysen ein sehr grosses Unfallvermeidungspotenzial fuer FAS im Bereich von ueber 70 Prozent aller schweren Unfaelle. Allerdings sind die Anforderungen an diese Systeme gross. Teilweise sind eine aktive Unterstuetzung oder ein Eingriff notwendig, was rechtliche und Akzeptanzprobleme mit sich bringt. Mit der dargestellten Methode der In-Depth Unfallanalyse aufgrund von Unfallprotokollen ist insgesamt eine Abschaetzung des Sicherheitspotenzials von Assistenzsystemen moeglich, wobei die Aussagen vorsichtig zu interpretieren sind. (Author/publisher) Bericht zum Forschungsprojekt 82.214/2001 (ITRD-Nummer D707790) der Bundesanstalt fuer Strassenwesen. English title: Drawing up requirements for driver assistance systems from the perspective of traffic safety. English abstract: The objective of the research, assigned by BASt and additionally promoted by the state of Lower Saxony, is, to draw up requirements for driver assistance systems (FAS) using the In-Depth-Analyses of accident data of the police in Braunschweig from the year 2002 and to estimate the advantages for safety that the introduction of such systems would provide. The false actions preceding the accident and their causes were analysed in the In-Depth-Analyses based on the accident forms. False actions are actions that lead to accidents. An FAS would have to correct these to prevent the accident, so that this analysis results in requirements to functionality (e.g. selection of a safe speed). The analysis of the causes of the false actions took place in accordance with an information processing model of human behaviour and provides references to the intervention strategy of the FAS. lf the cause is, for example, a lack of perception of existing information, a warning can prevent the accident. In the case of a false decision, active support from an FAS is necessary. False interpretations and errors in execution were also observed. The database for this project are 4,258 accident forms from Braunschweig from the year 2002 and 185,004 accidents in Germany that were made available by the Federal Office for Statistics as a 50% random sample of the official accident statistics of the year 2002. Accidents were selected from both data sources where the party responsible for the accident was a car and the driver (when known) was at least 18 years old. The serious accidents in Braunschweig (993 accidents) were ranked in such a way that they could be compared with the accidents in the rest of Germany with respect to the type of accident, day of the week and time of day. The accident forms were analysed and put into groups in the In-Depth-Analyses for 6 types of accident and for the most frequent sub-types with respect to false actions and causes. Agreements amongst assessors was checked in a process similar to random sampling and proved to be very satisfactory (Spearman rho = 0.91). The analysis shows three large groups of accidents from which three types of need for support can be derived. In the case of accidents occurring when turning or crossing, other road users are neglected during planning for various reasons, which can be described as a lack of perception of existing information. Seldom do false decisions play a role. Assistance while crossing, that can recognise vehicles that have priority from the right, the left or are oncoming and cyclists coming from the right, is required to prevent these accidents. 21,2 % of all serious accidents could be prevented by assistance while crossing that warns people about these vehicles. In the case of accidents while driving, lack of adaptation of speed to the road conditions, to the state of the driver and to ones own performance is predominant and can be described as a false decision. lt is partly the neglect of horizontal steering without any reason (lack of perception) that adds to this. Active support of speed regulations dependent on the situation with additional support of horizontal steering could prevent 20.4% of all serious accidents. Tailgating occurs in longitudinal traffic in particular, but also in a series of other accident types. A system of avoiding collisions with regulation of intervals and speeds dependent on the situation is necessary here, that can also recognise stationary vehicles and that supports braking in the case of sudden interventions. If the System were to fulfil these requirements, 32.5% of all serious accidents could be prevented. Since the most frequent causes for false actions He in the area of false decisions, active support of speed maintenance is required for this purpose. On the whole the analysis shows that there is much potential for avoiding accidents using FAS in the range above 70% of all serious accidents. However, there are immense demands to these systems. They have to recognise all vehicles that have priority coming from all directions in the area of the crossing, as well as characteristics of the situation such as the state of the road, obstacles on the carriageway and characteristics of the driver such as an impaired state. Active support on intervention is partly required, which results in legal problems or problems of acceptance. Using the method of in-depth analysis of accidents based on accident forms an estimate of the potential for safety of assistance systems is possible, on the whole, however, the statements must be interpreted with caution. (Author/publisher)
Samenvatting