Uncertainty, Efficiency, and Stability of Mixed Traffic Flow: Stochastic Model-Based Analyses

Abstract

This paper proposes a stochastic model for mixed traffic consisting of human-driven vehicles (HVs), connected automated vehicles (CAVs), and degraded connected automated vehicles (DCAVs). The model addresses the issue that most of the current literature ignores: the degradation of CAVs, and the heterogeneity and uncertainty of HVs, CAVs, and DCAVs. The source of uncertainty was the heterogeneous behavior of HVs, CAVs, and DCAVs, captured using vehicle-specific car-following relations, that is, parametric uncertainty. The proposed model allowed for the explicit investigation of the uncertainty, efficiency, and stability of mixed traffic under various CAV penetration rates, different positions of CAVs in the traffic stream, and the different degradation levels of CAVs. The numerical experiment results showed that a larger CAV penetration rate helped to reduce uncertainty and improve the efficiency and stability of traffic flow. Furthermore, we investigated the impact of different position combinations of CAVs in the mixed traffic stream on traffic performance under four scenarios: 1) CAVs randomly distributed in the traffic stream, 2) CAVs forming a platoon traveling in the front of the traffic stream, 3) CAVs forming a platoon traveling in the middle of the traffic stream, and 4) CAVs forming a platoon traveling in the rear of the traffic stream. The results demonstrated that Scenario 2 gave the best performance in reducing uncertainty and improving efficiency and stability under different CAV penetration rates, whereas Scenario 4 performed the worst. Moreover, increasing degradation levels of CAVs negatively affected the reduction of uncertainty and improvement of efficiency and stability.