A Review of Intelligent Systems for Driving Risk Assessment

Driving risk assessment is important to guide the actions, states and behaviours of drivers for the prevention of road incidents or accidents. With the widespread of sensors constantly gathering data about drivers' actions, vehicle position and location, vehicle characteristics and environmental conditions; artificial intelligence is now used to produce driving risk assessments. However, driving is a complex, uncertain and dynamic activity affected by the synergy and interaction between driver behaviour, vehicle position and external factors. In addition, driver behaviour is a multifaceted concept consisting of different personal traits, such as individual drivers' physical and mental states, drivers' responses, and driving styles. This paper reviews the different computational and artificial intelligence methods that have been investigated in the literature to characterise and assess the risks of driving behaviours using vehicle sensor data. The purpose of this review is to present the opportunities and gaps of existing intelligent methods in driver characterisation and driving risk assessment to inspire and guide future research in this area. It also aims to present opportunities of available and new data sources for the improvement of driving risk assessment.