Generative AI and Online Learning Based Road Rage and Aggressive Driving Detection

Usage-Based Insurance (UBI) revolutionizes traditional coverage by utilizing technology to tailor insurance premiums to individual driving behaviour. Road rage and Aggressive driving behaviour Detection (RAD) is pivotal for enhancing the precision of risk assessment and premium customization. This paper introduces a novel approach to UBI based on RAD using Global Positioning Systems (GPS) signals and heart rate monitoring through smartphones and wearable technology. To mitigate the skewed nature of real-world driving data, it employs the Conditional Tabular Generative Adversarial Network (CTGAN), while an online learning paradigm ensures the model remains dynamic and scalable, adapting to new data over time. The proposed system called the Incremental Learning-based UBI System with Temporal Analysis and Recognition (ILUBISTAR) introduces a transformative approach to UBI pricing by leveraging RAD. This research employs a Bi-directional Long Short-Term Memory (BiLSTM) network for processing sequential driving data acquired real-time to classify driving behaviour. This model is trained and evaluated with data acquired from a cohort, demonstrating an accuracy of 99.9% in identifying drivers with consistent bad driving behaviours. It also features desirable performance metrics in the detection of good, unhealthy, always bad and road rage & aggressive drivers. This research demonstrates ILUBISTAR’s effectiveness in identifying various driving behaviour classification, underscoring its innovative contribution to UBI pricing. This research not only showcases ILUBISTAR’s superior performance in identifying diverse driving patterns but also highlights its potential to revolutionize UBI by promoting safer driving and more equitable insurance premiums.