Bounding the efficiency of vehicle automation in general transportation networks

Emerging autonomous vehicle (AV) technology is expected to bring substantive benefits to the transportation systems. This study focuses on the improvement of travel efficiency in transportation networks with AVs. Inspired by the concept of price of anarchy (PoA), we introduce a novel indicator, the benefit of vehicle automation (BVA), to quantify the efficiency benefit brought by AVs. The BVA is defined as the ratio of the transportation network performances under the two user equilibrium states before and after the penetration of AVs. In characterizing the BVA, both the AV market penetration rates and the automation levels are used to describe AV deployment, and the impact of AVs in terms of capacity improvement and value-of-travel-time (VOTT) compensation are explicitly considered. Under mild assumptions, we derive a family of bounds of BVA that are independent of the network topology, including two types of lower bounds and four types of upper bounds. These bounds have different degrees of generality and effectiveness in distinct scenarios. Properties including tightness, monotonicity and generality of the bounds are discussed. Numerical experiments on general networks are conducted to verify the effectiveness and applicability of them. Furthermore, we also propose two bicriteria bounds of BVA. The first specifies the maximum level of rising demand whose induced efficiency loss can be offset by the benefit of AV penetration, while the second provides an upper bound on the worse-off degree of the performance due to non-uniform AV penetration across origin-destination pairs. This work is expected to deepen our understanding of the efficiency benefit of AVs in general transportation networks, and to provide a guideline for policymakers to navigate the integration of AVs.