Cloud-Edge Cooperative Distributed MPC With Event-Triggered Switching Strategy for Heterogeneous Vehicle Platoon
Jun-Xiao Zhao, Yaling Ma, Li Dai, Zhongqi Sun, Yuanqing Xia
Abstract
This paper proposes a cloud-edge cooperative scheme for heterogeneous vehicle platoon based on distributed model predictive control (DMPC) and event-triggered mechanism. In vehicle platooning, cooperative driving necessitates intricate vehicle-to-vehicle (V2V) interactions, which are governed by challenges including large-scale, constraints, nonlinearity, and stringent real-time processing. DMPC can decompose a high-dimensional, complex centralized optimization problem into multiple subproblems solved in parallel. This property can reduce inter-vehicle communication while enhancing the efficiency of control problem-solving, making it well-suited for platoon control scenarios. Considering the computational limitations of onboard devices, we design and deploy a distributed nonlinear model predictive control (DNMPC) algorithm on cloud servers with superior distributed computing capabilities utilizing docker containerization technology. However, this cloud-centric approach hinges on reliable network quality. To ensure the safety of platoon driving under network anomalies, we introduce edge computing, designing and deploying a less complex distributed linear model predictive control (DLMPC) algorithm at the vehicle edge. Moreover, we incorporate an event-triggered mechanism to optimize cloud computing utilization and reduce operational costs while maintaining control performance. Finally, the efficacy of the architecture is validated through a simulation of a heterogeneous vehicle platoon.