A Stackelberg Game Theory-Based Multi-Vehicle Decision Planning Model for Bus Exiting at Bus Bay Stops
Аннотация
Bus departure from bay stops often involves simultaneous interactions with several vehicles in the merging and adjacent lanes. This study develops a multi-vehicle bi-level Stackelberg decision-planning model for connected and automated traffic. The bus acts as the upper-level leader and selects its departure strategy by considering the combined payoff of the bus and surrounding vehicle group, while five follower vehicles determine individual longitudinal and lane-changing responses through a lower-level game. Safety, comfort, and travel efficiency are incorporated into the payoff functions, and the resulting decisions are mapped to feasible quintic-polynomial trajectories. NSGA-II is used for the upper-level search and sequential best responses are used for the lower-level game. Numerical experiments show that incorporating surrounding-vehicle payoff in the upper-level decision improves aggregate system payoff relative to a bus-oriented criterion. When higher adjacent-lane speeds change the selected lane-changing follower, the coordinated decision also maintains a substantially larger dynamic safety margin. The results demonstrate the value of considering multi-vehicle interactions in connected bus-departure control.