This study investigates personalized feedback mechanisms aimed at helping drivers manage their emotions to enhance road safety. Conducted in a simulated driving environment, the research involved 21 participants experiencing three distinct emotional states while driving: happy, neutral, and angry.

Our driving simulator integrates three feedback channels: a visual feedback system using LED lights, auditory feedback through built-in audio, and vibrotactile feedback via seat vibrations. Additionally, we collected facial landmark data using non-intrusive recognition technology for subsequent machine learning analysis.

Our motivation arises from the need to create emotion-based feedback systems for vehicles that surpass traditional generic approaches. Using a simulator setup, we examined the effectiveness of various modalities in regulating driver emotions across recorded emotional states. Unlike existing studies that mainly explore single feedback modalities or concentrate solely on emotion detection, our research investigates the combined effects of multiple feedback types. Our technical setup involved collecting facial landmark data through facial tracking and testing different combinations of feedback activations, focusing specifically on how visual, auditory, and vibrotactile feedback influence driver emotions.

Our analysis revealed that participants generally reported positive experiences with the feedback system, with preferences differing according to their emotional state. The facial landmark data gathered during the study was subsequently used to develop a machine learning-based emotion classification system, highlighting the potential for future real-time emotion detection in vehicles. For more comprehensive details on the studies and findings, please refer to our paper.

---

---

Team

Kevin Fred Mwaita, Antonella De Angeli, Michael Haller