The paper analysed driver intention models in context of partly automated vehicles and a conflict prevention strategy. Therefore two models, a driver and situation adaptive model and a model based on an average driver were implemented. Thereby the driver and situation adaptation was achieved by using fuzzy logic with adaptive fuzzy sets and rules. The fuzzy sets were trained offline during 90 seconds, which is quite short. The benefit and necessity of the adaptive intention model compared with the non-adaptive model could be shown. Even though the learnability of the non-adaptive model was quite easy for nearly all subjects through the activity display. This was not sufficient. The model had to be adaptive to obtain very good results. In the next step the conflict detection will be further improved. The goal is to extend the conflict detection by classifying the reason for the conflict. By knowing the reason for the conflict this allows to adapt the automation behaviour on the manoeuvre and/or trajectory level to the driver. This would be another strategy of conflict prevention. The advantage would be to reduce the number of transitions and thus to increase the acceptance on partly automated vehicles. Therefore an observer model will be implemented to learn the driver behaviour.
Driver intention modelling for partly automated vehicles - The benefit and necessity of a driver and situation adaptive approach
2012
7 Seiten, 3 Bilder, 4 Tabellen, 6 Quellen
Conference paper
English
DRIVER INTENTION IDENTIFICATION METHOD AND DRIVER INTENTION IDENTIFICATION APPARATUS
European Patent Office | 2018
|Modelling the Dynamics of Driver Situation Awareness in Automated Driving
Springer Verlag | 2017
|From driver to supervisor - an analysis of partly automated driving
Automotive engineering | 2013
|