AutoFlock - Two-Dimensional Distributed Control of Heterogeneous Traffic Flow with Flocking Cooperative Automated Vehicles

• Applicant: Professor Dr. Meng Wang
• Subject Area: Traffic and Transport Systems, Intelligent and Automated Traffic
• Term: since 2024
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 544900900

Traffic jams threaten our society. Cooperative automated vehicles (CAVs) revolutionize the way we control our vehicles today but bring tremendous changes and consequently uncertainties to congestion formation and propagation characteristics known to traffic scientists. These changes and uncertainties stem from the highly nonlinear coupling between microscopic vehicle behaviors and macroscopic traffic flow and pose serious challenges to the control of future mixed traffic with the coexistence of human-driven vehicles (HVs) and CAVs. Existing microscopic vehicle control approaches are myopic in space and time and ignore global flow dynamics in the decision-making of CAVs. Current macroscopic traffic control systems are centralized and suffer from scalability restrictions, subsystem failures and performance saturation. Critical scientific foundations for modelling and control of mixed HV-CAV flow are lacking. Inspired by biological groups in nature, the project aims to develop a new traffic control paradigm by reconstructing the fundamental building blocks of traffic flow with the novel concept of CAVs as two-dimensional (2-D) flocks. CAV flocks will collaborate with each other based on local communication and control mixed flows without giving direct control instructions to HVs. In particular, we will develop: 1) flow stabilizing strategies to prevent traffic flow congestion formation under linear disturbances; 2) wave absorbing strategies to eliminate nonlinear congestion waves. The control strategies will be built on a mesoscopic modelling approach of Smoothed Particle Hydrodynamics (SPH). This meshfree Lagrangian modelling method naturally describes communication topology, finite perception range and motion constraints of different vehicle classes at the microscopic level while unravelling macroscopic flow patterns in 2-D. It will enable the control of flow dynamics for up to 10 minutes, far beyond the reach of the existing CAV control paradigm with a prediction horizon of less than 10 seconds. The project will create a new paradigm for vehicular flow modelling and control. It will push the envelope of roadway capacity to a new level and establish scientific tools to answer big-picture questions at the macroscopic flow level enabled by revolutionary technologies before they enter the market.

• DFG Programme: Research Grants