The process of buses entering and exiting a station often accompanies mandatory lane change, which can significantly impact traffic safety and efficiency. However, current research on lane change strategies for intelligent and connected buses relies on single-vehicle algorithms, lacking consideration for surrounding vehicles and facing challenges in handling inter-vehicle conflicts. These approaches also struggle to ensure the safety and success rate of mandatory lane change during entry and exit, and easily impact the traffic flow. To improve the performance of mandatory lane change for intelligent and connected buses during the entry and exit processes, a two-vehicle cooperative lane change strategy based on the cloud control system (CCS) is proposed, which can schedule buses and cooperating vehicles simultaneously.
This study designed a set of cooperative lane change strategies for the entry and exit processes. For the entry process, the road segment leading to the bus station was divided according to the distance to the station, and a decision-making model based on the lane change benefit criterion was established. A two-stage trajectory planning method was developed for lane change trajectories, incorporating the longitudinal adjustment and cooperative lane change stages. Optimization problems were formulated and solved in a receding horizon to obtain the optimal trajectories. A layered quadratic programming method was also employed to improve the real-time performance of the algorithm. For the exit process, to adapt to the characteristics of the exiting motion of buses, a rule-based decision-making method and pre-deceleration rule were designed. Lane change trajectories were optimized considering the characteristics of bay-style bus stations to meet the station shape and bus mobility requirements. Finally, the NGSIM dataset was used to design typical scenarios for simulation and hardware-in-the-loop experiments. The effectiveness of the proposed strategy was tested. The proposed strategy was also compared with baseline methods, i.e., the minimizing overall braking induced by lane change (MOBIL) decision model along with the fifth-order polynomial single-vehicle planning method, and the rule-based decision model along with the fifth-order polynomial single-vehicle planning method for the entry and exit processes, respectively.
The simulation and hardware-in-the-loop experiments under typical scenarios demonstrate that the proposed strategy can effectively adapt to realistic vehicle speed changes, assisting buses in completing lane changes during station entry and exit. Moreover, when faced with the actual computing and communication environment of the cloud platform, the proposed strategy meets the requirements for real-time and effective calculations. The comparison with the baseline method shows that, in batch tests for station entry and exit, the lane change success rates of the proposed strategy were 87.4% and 82.4%, respectively, both higher than the baseline method's 51.9% and 20.4%, respectively. Evaluation metrics based on the following vehicle's speed, acceleration, and time to collision (TTC) also indicate that, compared with the baseline method, the proposed strategy can reduce the impact of mandatory lane change on traffic efficiency while maintaining better safety standards.
Using the CCS to achieve centralized decision-making and trajectory planning for scheduling buses and cooperating vehicles simultaneously, the proposed strategy can effectively handle mandatory lane change under high traffic density conditions. The proposed strategy not only improves the success rate of mandatory lane changes for buses entering and exiting stations, but also ensures lane change safety and reduces the impact on upstream traffic. The proposed strategy is also applicable in cloud platforms under realistic computing and communication environments.