Optimal Formation of Autonomous Vehicles in Mixed Traffic Flow

Keqiang Li,Jiawei Wang,Yang Zheng
DOI: https://doi.org/10.1016/j.ifacol.2020.12.2301
2020-04-01
Abstract:Platooning of multiple autonomous vehicles has attracted significant attention in both academia and industry. Despite its great potential, platooning is not the only choice for the formation of autonomous vehicles in mixed traffic flow, where autonomous vehicles and human-driven vehicles (HDVs) coexist. In this paper, we investigate the optimal formation of autonomous vehicles that can achieve an optimal system-wide performance in mixed traffic flow. Specifically, we consider the optimal $\mathcal{H}_2$ performance of the entire traffic flow, reflecting the potential of autonomous vehicles in mitigating traffic perturbations. Then, we formulate the optimal formation problem as a set function optimization problem. Numerical results reveal two predominant optimal formations: uniform distribution and platoon formation, depending on traffic parameters. In addition, we show that 1) the prevailing platoon formation is not always the optimal choice; 2) platoon formation might be the worst choice when HDVs have a poor string stability behavior. These results suggest more opportunities for the formation of autonomous vehicles, beyond platooning, in mixed traffic flow.
Optimization and Control,Systems and Control
What problem does this paper attempt to address?
The problem that this paper attempts to solve is how to find an optimal formation form of autonomous vehicles in mixed traffic flow to achieve the best performance of the entire traffic system. Specifically, the paper focuses on the optimal formation method of autonomous vehicles in the mixed traffic flow containing human - driven vehicles (HDVs) and autonomous vehicles (AVs). The paper considers the optimal \(H_2\) performance of the entire traffic flow, which reflects the potential of autonomous vehicles in alleviating traffic disturbances. By modeling the optimal formation problem as a set - function optimization problem, the research discovers two main optimal formation forms: uniform distribution and platoon formation, depending on traffic parameters. Moreover, the research also points out that platoon formation is not always the optimal choice; when the serial stability behavior of human - driven vehicles is poor, platoon formation may be the worst choice. These results indicate that in mixed traffic flow, autonomous vehicles have more formation opportunities besides platoon formation.