Abstract:
Road rage has become one of the most significant issues in society, risking the safety of drivers, passengers and pedestrians. Road accidents, physical confrontations and even fatalities can occur due to road rage. This research is focused on creating a real-time road rage detection system using only smartphones and smartwatches to help the drivers to mitigate possible road rage incidents.
The system employs a multimodal fusion approach by integrating data from facial emotion recognition, accelerometer readings, and heart rate signals. A Support Vector Machine (SVM) classifier is used for facial emotion recognition to distinguish between angry and non-angry states. For driving behaviour classification, the y-axis accelerometer data is analysed to identify sudden accelerations and braking events, using a Random Forest (RF) model. Heart rate analysis is performed through a rule-based algorithm to detect physiological indicators of rage. The binary outputs from all three modalities are then combined using a weighted fusion formula to calculate a rage score. If this score exceeds a predefined threshold, the system flags it as a road rage event and issues a real-time alert to the drivers.
The facial emotion recognition model achieved an accuracy of 76%, while the driving behaviour classification model reached 80% accuracy. Both models demonstrated reliable performance within acceptable margins, validating the effectiveness of the proposed system.
