Driving Safety Risk Analysis and Assessment in a Mixed Driving Environment of Connected and Non-Connected Vehicles: A Systematic Survey

With the continuous development of intelligent networks and autonomous driving technologies, heterogeneous traffic flow represented by conventional vehicles (CV), autonomous vehicles (AV), and connected and autonomous vehicles (CAV) have emerged, and consequently, the driving safety risk issues in this mixed driving environment have become increasingly complex. In a mixed and connected environment, different traffic streams in complex driving scenarios are intertwined with each other, and driving behaviours such as steering and lane-changing between various traffic streams are frequent, thus increasing the crash risk of vehicles. To understand the research methods, theoretical models, and system architectures in the field of driving safety analysis in this mixed-connected environment, this article reviews the driving safety risk progress from four major aspects: driving safety risk perception and identification, driving safety risk prediction, driving safety risk quantification, and driving safety risk early warning. By summarizing the existing research, it can be found that academics have achieved many achievements in urban driving safety risk evaluation in mixed-connected environments. Still, there are several problems and challenges that need to be solved, such as the stability and reliability of collaborative sensing systems of AV and CAV in complex traffic environments, the accuracy of object detection, the vehicle information security, the limitations of a single factor analysis used in driving safety risk assessment, and the accuracy of trajectory prediction for both the CV, AV, and CAV. By analyzing the limitations of the existing research, this article proposes a future research direction, which provides a reference for the development of driving safety risk research.