Effect of a dynamic cooperative lane-changing algorithm on traffic evolution of connected and automated vehicles system at diverging on two-lane highways

The diverging sections of two-lane highways are a typical traffic bottleneck in urban cities. Due to unregulated lane-changing (LC) behaviors, velocity oscillation and traffic congestion frequently occur. In this paper, a dynamic cooperative lane-changing (DCLC) algorithm is introduced to coordinate the LC behaviors of connected autonomous vehicles (CAVs) in a control zone upstream of the diverging point. For a CAV with LC demand, its virtual leader and follower on the adjacent lane are dynamically determined based on relative distances in real time. It will change lane if the safety time gap is satisfied. With the normal lane-changing algorithm as a benchmark, the phase transition and evolution of the traffic system are investigated and compared. Simulation results demonstrate that the proposed DCLC algorithm has a positive effect on traffic capacity. After the DCLC algorithm is implemented, the LC behaviors of CAVs are concentrated in the middle of the control zone, rather than being distributed near the diverging point in the benchmark. Moreover, some system characteristics have also been improved significantly. Finally, the sensitivities of demand and safety time gap in LC are analyzed to verify the universality of the DCLC algorithm in improving traffic performance.