This paper is committed to capturing the dynamic behaviors of homogenous flow of autonomous vehicles (AVs), and exploring the control strategies to improve traffic conditions, which can alleviate traffic congestion and improve traffic efficiency. Firstly, a car-following model of AVs considering real-time driving state is established. Secondly, based on bifurcation theory and stability theory, bifurcation analysis is carried out and the relationship between bifurcation and stability is revealed. In order to suppress the bifurcation and improve the stability, a controller considering multi-step prediction and memory mechanism (MPM) is designed, and the root trajectories for eigenvalues and stable time length of the model controlled by MPM controller are calculated. In response to the limitations of the MPM controller, a hybrid controller including the MPM controller and PID controller is further proposed, and it is found that the model controlled by hybrid controller has greater range of stable bifurcation parameter and stable time length, which means better ability of bifurcation suppression. Finally, the capabilities of the controller proposed in this paper are effectively demonstrated by numerical experiments in MATLAB and simulation experiments in the ROS-Gazebo environment.

中文翻译：

---

通过多步预测和记忆机制与PID相结合的混合控制策略实现自动驾驶车辆的交通流分叉控制


本文致力于捕捉自动驾驶汽车（AV）同质流的动态行为，探索改善交通状况的控制策略，从而缓解交通拥堵，提高交通效率。首先，建立考虑实时驾驶状态的自动驾驶汽车跟车模型。其次，基于分岔理论和稳定性理论，进行分岔分析，揭示分岔与稳定性之间的关系。为了抑制分岔、提高稳定性，设计了考虑多步预测和记忆机制（MPM）的控制器，并计算了MPM控制器控制的模型特征值和稳定时间长度的根轨迹。针对MPM控制器的局限性，进一步提出了包含MPM控制器和PID控制器的混合控制器，发现混合控制器控制的模型具有更大的稳定分岔参数范围和稳定时间长度，意味着更好的稳定性。分叉抑制能力。最后，通过MATLAB中的数值实验和ROS-Gazebo环境中的仿真实验，有效地证明了本文提出的控制器的能力。