Vehicle intelligence for mobility: A simulation study of mixed traffic dynamics

With the advancement of automatic control and artificial intelligence, the individual intelligence of autonomous vehicles (AVs) has evolved significantly. The vehicle intelligence not only enhances perception and decision-making but also drives AVs to pursue the optimization of individual mobility through selfish and aggressive lane-changing maneuvers. To investigate the impact of such behaviors on mixed traffic, this paper systematically analyzes the mechanism by which AV lane-changing aggressiveness affects traffic flow performance. First, a three-lane highway simulation scenario is constructed based on the SUMO microscopic traffic simulation platform, modeling the heterogeneity of the mixed traffic flow. Subsequently, an evaluation metric system covering mobility and comfort is established, where average velocity is adopted as the direct quantitative proxy for mobility (i.e., traffic efficiency). Finally, simulation experiments are designed to quantify the dynamic impacts of AV selfish lane-changing behaviors. The results indicate that (1) the impact of AV selfish lane-changing on mixed traffic performance exhibits non-linear characteristics and is significantly correlated with traffic volume and AV penetration rates. (2) In scenarios characterized by low traffic volume and moderate AV penetration, a moderately selfish lane-changing strategy can effectively act as a mobility enabler, increasing the average velocity of AVs by up to 3.89%; however, excessively aggressive maneuvers can destabilize the traffic flow, precipitating a significant deterioration in the overall mobility of the mixed traffic. (3) A critical trade-off mechanism exists in the pursuit of mobility: gains in speed inevitably incur a penalty in passenger comfort. Therefore, future strategy designs must carefully balance mobility gains against comfort losses. The findings of this study provide theoretical support for the design of AV microscopic behavior control and mixed traffic management.