Surrogate Safety Measures: Review and Assessment in Real-World Mixed Traditional and Autonomous Vehicle Platoons

Abstract

Surrogate safety measures (SSMs) are critical tools for evaluating the safety performance of mixed traffic. Crashes are rare events, and historical crash data is scarce for mixed traffic that includes autonomous and/ or connected vehicles. Recent safety review papers focus on traditional human-driven vehicles (TVs) and do not encompass advanced technology vehicles such as Autonomous Vehicles (AVs), Connected Vehicles (CVs), and Connected-Autonomous Vehicles (CAVs). This study examines the development, implementation, and shortcomings of SSMs, and SSM-based models used for mixed traffic safety evaluation. It reviews the current relevant literature and applies a case study analysis using a real-world mixed traffic dataset. The study summarizes the fundamental SSM guiding concepts, as well as their most significant metrics including threshold values employed in the past for SSMs and SSM-based models. Primary benefits and limitations of examined SSMs and SSM-based models are also underlined. This review reveals significant gaps in the literature that might guide future research paths in SSM-based mixed traffic safety assessment. Critical gaps include the absence of robust SSM threshold selection criteria, the suitability of current SSMs in mixed traffic research, microsimulation modeling that lacks proper calibration and validation, and the absence of a framework for selecting or combining multiple SSMs.INDEX TERMS Mixed traffic, surrogate safety measures, real-world mixed traffic dataset, autonomous vehicles, connected vehicles, connected-autonomous vehicles

Control Transitions in Level 3 Automation: Safety Implications in Mixed-Autonomy Traffic

“…A good categorization, including mathematical definitions of relevant SSMs, can be found in a survey in [35]. Another comprehensive and quite deliberate literature review is given in [36], which details state-of-the-art SSMs in mixed-autonomy traffic research. Particularly, the authors critically discuss certain shortcomings of SSMbased safety assessments and point out inadequacies in simulation-based safety evaluations, which concentrate on automated driving without calibrated vehicle models.…”

“…For an energy-based SSM, we reference [39] as an example, which presents an interesting, novel approach called 'extended delta-V', which estimates the crash severity of potential conflicts. Since this work does not delve deeply into SSM-based models or other related approaches, we refer the interested reader to [35,36] for more comprehensive information. However, we want to allude to a few notable papers that helped us develop a feasible approach for our own case study (see Section 3.3).…”

“…As pointed out by [35,36], the absence of human perception-reaction times (PRT) or response times (RT) is a noted concern. In response to this issue, we introduce the Modified Deceleration Rate to Avoid Crash (MDRAC), which was developed by [43] as an enhancement over the traditional DRAC.…”

Coalitional Control Strategy for a Heterogeneous Platoon Application

“…According to the review work in [15], there are three important research challenges to be further addressed within the vehicle platooning research community: (i) developing adequate theoretical models to describe a vehicle platoon, (ii) testing and real-time experiments for the platooning algorithms to validate the strategy, and (iii) introducing security solutions in the platooning algorithms, to counteract possible security attacks. The safety in vehicle platooning is a valid concern tackled in the review works [16,17]. The string stability of a platoon in the presence of communication delays and proposed compensation strategies available in the literature are reviewed in [18].…”

Identifying the Threshold Discrepancy of Rear-End Conflicts under Clear and Rainy Weather Conditions Using Trajectory Data

“…provided the assumptions 22 . Only very few previous studies begun to recognize that the selection of thresholds should consider specific traffic characteristics such as road user types, geometric types and mixed traffic flows.…”