Evaluating and enhancing the safety performance of automated longitudinal control at on-ramp merging bottleneck: A simulation study in the framework of Kerner’s three-phase traffic theory

The adaptive cruise control (ACC) system, an essential component of commercial autonomous driving that functions in longitudinal control, has attracted significant research interest because of its potential to reduce accident rates. By introducing a non-fixed headway concept, Kerner recently proposed the Three-traffic-Phase ACC (TPACC) model, which integrates a speed adaptation module derived from the three-phase traffic theory. However, the efficacy of such headway flexibility in safety performance remains underexplored. To fill this gap, this study aims to evaluate the longitudinal crash risk of TPACC at on-ramp merging bottleneck, a site where complicated traffic dynamics often constitutes concern, and to further bolster its safety performance. Specifically, simulations are conducted to evaluate the longitudinal crash risk of pure TPACC traffic as well as human-machine mixed traffic, by comparing it with a classical ACC model across various flow rates and market penetration rates. Enhancement of TPACC's safety performance is achieved through the multi-objective optimization in regard to a parameter set of the speed adaptation module. The results indicate that TPACC has the ability to weaken disturbances caused by merging vehicles and leads to a markedly decline in the longitudinal crash risk. Mixed traffic incorporating with TPACC vehicles also exhibits superior safety performance over that with the classical ACC vehicles. Furthermore, the proposed optimization measure effectively enhances TPACC’s longitudinal safety performance across a vast spectrum of penetration rates and flow rates at on-ramp merging area, all the while preserving throughput efficiency. On this basis, a guideline for directional tuning of the parameter set under conditions of limited information accuracy is provided, thereby heightening its practicality.