Application of machine learning techniques for driving errors analysis: systematic literature review

Driving is a challenging process, any misinterpretation or interaction error during this process can result in an accident, as such, leading to the concept of driving errors (DE), one of the largest concepts explored in the literature. Yet, it is still not understood. This research aims to systematically analyse DE by providing a unique definition of this concept, examining the most common errors during driving, and providing insight into the application of machine learning (ML) algorithms in this context. Thus, a systematic literature review (SLR) of empirical studies published over the past decade (2010-2020) was conducted. 75 primary studies were identified as relevant to the purpose of this research and potentially useful for a thorough understanding of DE. In the beginning, we provide a comprehensive definition of DE, followed by an overview of the literature on the application of ML in DE. The results show that the most popular types of errors behind the well are decision-making errors and recognition errors. Moreover, ML models have been widely used in recent years to analyse DE, especially those based on artificial neural networks or regression. Furthermore, the majority of these models are based on real data extracted using non-embedded instruments or questionnaires.