Current Trends in Driver Safety Monitoring: Datasets, Prediction Tasks, and Machine Learning Approaches

Driver safety is a major research topic in intelligent transportation systems, where the ability to monitor drivers and their surroundings can support the development of safer and more effective assistance systems. Within the context of a funded research project aimed at providing a system for driver and environment monitoring, this work presents a preliminary state-of-the-art analysis of the scientific literature. The study focuses on three main aspects particularly relevant to the project objectives: the datasets currently available, the prediction tasks previously addressed, and the machine learning algorithms used for driver related monitoring problems. The reviewed literature covers a wide range of safety-related tasks, including drowsiness and fatigue detection, distraction recognition, stress and cognitive state assessment, driving behavior analysis, and intention prediction. At the same time, the analysis highlights the variety of sensing modalities and methodological choices adopted in existing studies, ranging from physiological and behavioral signals to multimodal approaches integrating vehicle and contextual information. This paper aims to provide a structured overview of the current landscape, identify the resources and research directions most relevant to the project, and offer a useful reference for the broader research community interested in driver safety and monitoring. The paper concludes with a discussion of the main findings and of the aspects that deserve further attention in future research.