Do cut-ins matter: Assessing the impact of lane changing and string stability on traffic flow

In recent years, much emphasis has been placed on developing driving strategies for a small number of autonomous vehicles (AVs) or even adaptive cruise control (ACC) vehicles to stabilize traffic flow and reduce traffic oscillations. Many of these strategies rely on more passive car following behavior of the AV, or larger time gaps for the AV than the human-driven traffic. A common criticism of these driving strategies is that they encourage human drivers to cut in in front of the AV, which induces additional oscillations. However, until now, this claim has been largely untested. Specifically, it is unclear how the increase in oscillatory traffic conditions as a result of increased cut-ins compares to the increase that results from poor car following behavior. This study presents a simple, simulationbased analysis to answer this. A large vehicle trajectory database recorded in real traffic is analysed to understand the characteristics of typical cut-ins, and these cut-ins are simulated using common car following models for human-driven and ACC vehicles. We find that for typical cut-ins, poor driving behavior of human drivers increases oscillations by roughly 156%, while increased cut-ins increases oscillations by only 77%. Thus, we conclude that AV or ACC driving strategies that stabilize the traffic flow by engaging in more passive car following behavior than the more aggressive human drivers can still be beneficial, even if it induces additional cut-in maneuvers.