The modern intelligent transportation system opts for accident prediction modules as a critical aspect for road safety. Where, an accident is predicted before it actually happens and precautionary measures be taken for its avoidance. Accident prediction methods’ analysis is also popular for critical infrastructure protection. Recently, Deep Learning (DL) techniques coupled with the classical Artificial intelligence (AI) methods have produced promising results in various domains. This work presents a DL and AI-based system of identifying driving risks for the Light Transport Vehicles (LTVs) that generate early warnings before an anticipated accident. These warnings enable to avoid critical traffic accidents. The proposed system collects and learns various driver’s LTV driving style patterns and afterward applies DL to classify the driver in one of the predefined classes. The proposed system consists of a custom-built driving simulator integrated with a computer that provides real-life driving experience for data recording. The miniature hardware simulator mimics an LTV consisting of a steering wheel, clutch, brake accelerator, gear assembly, and indicators. The collected data is first evaluated using unsupervised learning methods. Afterwards, it is used to train the DL classifier. The obtained results show that the proposed system attains an average classification accuracy of 85%, where the maximum accuracy of 97% is observed for the gradient boosting classifier.