Skip to content

soc-ucsd/mixed-traffic

Repository files navigation

Modeling and Control of Mixed Traffic Flow

In this project, we investigate dynamical modeling and fundamental properties for ring-road mixed traffic systems with Connected and Autonomous vehicles (CAVs) and human-driven vehicles (HDVs). The control policies CAVs are designed with a system-level consideration in terms of

  1. dissipating traffic waves
  2. improving the performance of the entire mixed traffic flow.

In particular, we demonstrate three important topics in mixed traffic flow.

To run the code, the Modeling package, YALMIP, and the optimization solver, MOSEK, are needed to solve the semidefinite program in controller synthesis.

Topic 1: Smoothing Traffic Flow via Autonomous Vehicles

Although large-scale numerical simulations and small-scale experiments have shown promising results, a comprehensive theoretical understanding to smooth traffic flow via AVs is lacking. Here, from a control-theoretic perspective, we establish analytical results on the controllability, stabilizability, and reachability of a mixed traffic system consisting of HDVs and AVs in a ring road.

A schematic diagram for the mixed traffic system is as follows.

Two demonstrations are shown below:

All the vehicles are HDVs: traffic wave emerges

There is one AV: dissipating traffic wave

Reference

  • Zheng, Y., Wang, J., & Li, K. (2020). Smoothing traffic flow via control of autonomous vehicles. IEEE Internet of Things Journal, 7(5), 3882-3896.[pdf]

Topic 2: Structured Optimal Control of Autonomous Vehicles

Due to the limited communication abilities in practice, the CAV can only receive partial information on the global traffic system for its feedback. Therefore, it is important to consider the local available information of the neighboring vehicles. This leads to the notion of structured controller design.

Here is an illustration of structured constraints under limited communication abilities.

References

  • Wang, J., Zheng, Y., Xu, Q., Wang, J., & Li, K. (2020). Controllability Analysis and Optimal Control of Mixed Traffic Flow with Human-driven and Autonomous Vehicles. IEEE Transactions on Intelligent Transportation Systems, 1-15.[pdf]
  • Wang, J., Zheng, Y., Xu, Q., Wang, J., & Li, K. (2019, June). Controllability analysis and optimal controller synthesis of mixed traffic systems. In 2019 IEEE Intelligent Vehicles Symposium (IV) (pp. 1041-1047). IEEE. [pdf] [poster]

Topic 3: Cooperative Formation of Multiple Autonomous Vehicles: beyond platooning

In mixed traffic flow, the prevailing platooning of multiple AVs is not the only choice for cooperative formation. We re-design the control strategies of AVs in different formations and investigate the optimal formation of multiple AVs using a set-function optimization perspective. Two predominant optimal formations, i.e., uniform distribution and platoon formation, emerge from extensive numerical experiments.

Uniform Distribution

Platoon Formation

References

  • Li, K., Wang, J., & Zheng, Y. (2020). Cooperative Formation of Autonomous Vehicles in Mixed Traffic Flow: Beyond Platooning. arXiv preprint arXiv:2009.04254.[pdf]
  • Li, K., Wang, J., & Zheng, Y. (2020). Optimal Formation of Autonomous Vehicles in Mixed Traffic Flow. In 21st IFAC World Congress. [pdf] [slides]

Contacts

Related project: LCC (Leading Cruise Control).