AI-Based Driver State Detection

In the Context of Handover Situations of (Partially) Autonomous Driving

Project Justification

In the next decades, automated vehicles will become increasingly established. Nevertheless, drivers will remain involved in monitoring and hand over tasks. Therefore, a central element in automated driving is this ability to safely hand-over. This driver readiness refers to the driver’s ability to guide the vehicle appropriately in the relevant situation and is thus a central prerequisite for a successful and safe driver-vehicle interaction.

Possible experimental setup for measuring driver’s facial expressions, eyelid movements, posture, and seat position

Our Approach

The aim of DrAIve, a project funded by the Federal Ministry of Education and Research (BMBF), is to make driver readiness measurable “just in time”. For this purpose, an artificial intelligence-based driver assistance system is being developed which automatically classifies the driver’s characteristics and behaviors. More specifically, audio and video data are used to make predictions regarding the driver’s readiness. These predictions are based on facial expressions, eyelid movements, heart rate, voice, posture, and seat position. The system should offer the possibility of detecting critical driving conditions in a short time by analyzing both the driver and the vehicle interior for critical events. DrAIve can thus contribute to a reduction of critical high-risk situations in road traffic.

ixp test track in Datteln

Work Packages, Insights and Outcomes

Ixp is involved in leading the requirement analysis. ixp acquires the needs of users as well as the prerequisites for the future classification data and measurement systems. Additionally, ixp oversees corpus engineering through lab and field studies maintaining high data quality suitable for future analyses. Afterwards, ixp supports the project partners within the development of the audio-, video- and cabin activity-based driver state detection module with a focus on the training of deep-learning models and evaluation of first prototypes. Finally, ixp is responsible for functionality testing, usability studies, and subsequent revisions and fine-tuning to ensure the overall quality and success of the project.

Overview of the framework, hypotheses, and proposed approach of the DrAIve Project

Related Projects