Analyse von Aktivitäten eines Fahrzeugführers zur Verhaltensbeschreibung am Beispiel des Fahrstreifenwechsels

Advanced driver assistance systems in modern vehicles have contributed to the reduction of accidents and less traffic participants being injured or killed in recent years. The systems warn the driver of imminent danger or actively interfere with the control of the vehicle in critical situations. Nonetheless, there is concern, as it can happen that the driver could have averted the danger himself and thus feels patronized by the system or perceives the warning respectively the intervention as an error or as unnecessary, which results in a decline of acceptance for the system. If the driver subsequently disables the system completely the important assisting effect is lost. To avoid erroneous or unnecessary interventions of the assistance system, we can attempt to predict the driver’s short-term behaviour using driver intention recognition. As driving behaviour shows intraand interindividual differences and is generally described as very complex and variable, the field of driver behaviour and its prediction requires intensive research. A closer consideration reveals a specific research gap. There is no universal description method available that allows a precise analysis of the driver’s actual behaviour while he is driving. We can only determine approaches to describe the nominal behaviour but have only very limited use for behaviour analysis. In the following work a method to describe the drivers behaviour is developed. This will allow for a fundamentally better understanding of the driver and his behaviour while controlling a vehicle, in order to develop new findings for a prediction. The description method is developed using a deductive approach. This means that the description method is derived from theoretical, universal findings. This work will determine that the behaviour is described on the basis of individual action steps. Action steps should be understood as singular, measurable “elementary partial steps” of the behaviour. The central element of the description method is a catalogue of the action steps of a driver. The catalogue is comprised of a total of six classes (visual information intake, lateral guiding, longitudinal guiding, signalling, gear changes and additional action steps). From comparative literature using measured values, analysable properties and characteristics are derived. On the basis of the catalogue it is possible to describe the drivers behaviour and to identify the possible behavioural patterns that can belong to a specific manoeuvre. The description method, that also includes a possibility for visualisation, is then used on an example to determine its suitability concerning the analysis of the drivers actual behaviour. The example manoeuvre is going to be a lane change to the left, as the expected behaviour of this manoeuvre includes many aspects of the description method. Concerning the behaviour that can be expected during the lane change, hypothesis will be formulated on the basis of a literature research which will then be translated into the description language of the developed method. In order to create a data base for the exemplary application, driving tests were carried out in a controlled environment. The data of 167 lane changes of a total of 69 test subjects was collected and analysed concerning the action steps at hand. As a result of the hypothesis verification characteristic activities of the driving manoeuvres were determined, that can be very helpful for a behaviour prediction. It has become evident that fixations on the left side mirror, the interior mirror, the target lane as well as glances over the left shoulder, specific steering action steps and use of the direction indicator are characteristic for the exemplary manoeuvres. In addition, there is an analysis of the sequence and the time reference of the start of the manoeuvre, allowing for extended predictions on the driving behaviour during the manoeuvre, which could be relevant for intention recognition. This work will conclude with a discussion of the results of the driver behaviour that was examined during the exemplary application. In addition, the developed method will be discussed and the outlook will show where further research concerning the driver and his behaviour is necessary