F-RELACS – Recognising and counteracting frustration at the wheel
F-RELACS – Recognising and counteracting frustration at the wheel
The notoriously slow driver in the car in front is annoying, the traffic jam never ends - frustration increases and concentration wanes. In the F-RELACS project, we are investigating how an automated vehicle can help drivers in this situation.
Frustration encourages aggressive behaviour while driving and therefore poses a safety risk for the driver themselves and for other road users. The use of automated vehicles can also cause frustration. For example, if the driver cannot understand a certain automated driving manoeuvre, if the menu navigation in the cockpit is not self-explanatory or if they are not sure whether the vehicle is actually designed for certain situations. This subjective uncertainty can lead to frustration and thus to reduced acceptance of automated systems. However, acceptance is a prerequisite for the spread of automated vehicles and the associated hoped-for increase in efficiency and safety.
The F-RELACS project is therefore dedicated to the design of so-called emotion-sensitive systems that recognise the current level of frustration of vehicle occupants and offer user-centred assistance based on this. As part of F-RELACS, the DLR Institute of Transportation Systems is cooperating with TWT Science & Innovation GmbH from Stuttgart and SoundReply GmbH from Cologne.
The aim of F-RELACS is to develop and demonstrate such a frustration-sensitive system. The DLR Institute of Transportation Systems is building a frustration detection and reduction system together with the project participants.
To this end, the DLR Institute of Transportation Systems is working with TWT in simulator and real driving studies to develop a tool for determining the driver's current frustration level from video recordings of the face and physiological data (heart rate, pupil width). To this end, driving simulator studies are first carried out and algorithms for recognising frustration are developed on the basis of the data. In addition, the probable cause of the driver's frustration is narrowed down using artificial intelligence on the basis of previous inputs to the user interface or vehicle parameters. Once frustration and its source have been recognised, the system takes appropriate de-escalating measures, primarily based on voice interaction. Possible interaction strategies can include "active listening" or changes to the vehicle interior (e.g. scents, light colours). All three project participants in F-RELACS are working together to develop the de-escalating measures.
The project results therefore form the basis for the development of innovative, empathetic assistants that are able to recognise drivers' emotions and offer tailored support based on this.
