Modeling decentralized mandatory lane change for connected and autonomous vehicles: An analytical method

Mandatory lane change (MLC) is the core of lane-specific motion planning for connected and autonomous vehicles (CAVs). To model the MLC for CAVs, this paper presented a decentralized cooperative MLC (DC-MLC) framework. The framework aimed to achieve a safe MLC maneuver with multiple-CAV cooperation, including the lane change times, positions, and trajectories for the CAVs, while minimizing the impacts of the MLC maneuver on overall traffic. This problem was rigorously formulated and solved by an analytical method that significantly decreases computational time and renders the methodology suitable for practical implementations. To evaluate the performance of the proposed DC-MLC framework, we built a CAV testing environment by referring to the next generation simulation program (NGSIM) US 101 dataset. A series of numerical experiments were conducted under the built CAV testing environment. The results showed that the proposed DC-MLC framework could achieve safe, traffic-efficient, and comfortable MLC maneuvers. Moreover, the trajectories generated by the proposed DC-MLC framework outperformed those of human drivers in terms of speed standard deviation, maximum acceleration/deceleration, and maximum acceleration jerk.