Hier finden Sie wissenschaftliche Publikationen aus den Fraunhofer-Instituten.

Driver intention algorithm for pedestrian protection and automated emergency braking systems

: Diederichs, Frederik; Schüttke, Tobias; Spath, Dieter


Institute of Electrical and Electronics Engineers -IEEE-; IEEE Computer Society:
IEEE 18th International Conference on Intelligent Transportation Systems, ITSC 2015. Proceedings : Smart Mobility for Safety and Sustainability; 15-18 September 2015, Gran Canaria, Spain
Los Alamitos, Calif.: IEEE Computer Society Conference Publishing Services (CPS), 2015
ISBN: 978-1-4673-6595-6 (Print)
ISBN: 978-1-4673-6596-3
International Conference on Intelligent Transportation Systems (ITSC) <18, 2015, Gran Canaria>
Bundesministerium für Wirtschaft und Technologie BMWi
19S12009J; UR:BAN
Urbaner Raum: Benutzergerechte Assistenzsysteme und Netzmanagement
Fraunhofer IAO ()

Driver intention recognition reveals an enormous potential for automated driving. Current automated emergency braking (AEB) systems warn and take over vehicle control in case of frontal collision risks. It is of high interest to detect driver’s intention before initiating such transitions to automated driving in order to avoid annoyance. Also the number of warnings can be reduced when driver intention recognition indicates a safe behavior planning. Especially pedestrian protection systems will benefit in urban environments where pedestrians potentially enter the vehicle trajectory frequently. The paper describes the development of a driver intention detection algorithm for automated emergency braking systems. The data for the development is collected in a driving simulator study with 32 participants, balanced roughly by gender, age and driving style. Every participant circulated in an urban environment with random pedestrian traffic. In six scenarios data were collected where a pedestrian was either on the road or walked towards the road. An automated emergency braking system would activate a warning and eventually an automated braking in such situations. This could be adapted or cancelled in case of an existing braking intention by the driver. A fixed base Face Lab Eyetracking device stored gaze patterns in relation to the pedestrians’ position on the presentation screen. Pedal activity was also recorded. The analysis of both data sources allowed the development of a rule based algorithm to detect driver’s intention to brake due to the pedestrian. The algorithm proposes a sequential analysis of eye gaze and pedal activity and is presented and discussed in this paper. The algorithm correlated highly with a driver intention recognition reference method developed earlier by Diederichs & Pöhler [1]. In 94 % of the scenarios the algorithm and the reference method match in their results.