Heterogeneous Traffic Management Using METANET Model with Filtered Feedback Linearization Control Approach

This paper presents a Filtered-Feedback-Linearization (FFL) controller for a non-signalized heterogeneous traffic network consisting of human-driven and autonomous vehicles at a macroscopic level. The FFL controller requires limited model information and it is effective for command following and rejection of unknown-and-unmeasured disturbances. This paper distinguishes between human-driven and autonomous vehicles in terms of their operational characteristics and controllability. To describe the traffic network’s behavior, we introduce an extended heterogeneous METANET model wherein the traffic flow, density, and velocity dynamics of each vehicle class are described. To develop traffic control policies, we propose a filtered-feedback-linearization control approach wherein the autonomous vehicles and human-driven vehicles are modeled as controllable agents. FFL controller sends the optimal suggested velocity of autonomous vehicles and human-driven vehicles as the controller command to the traffic system to reach the desired velocity, density, and average flow rate of each sub-network. Numerical simulation demonstrates the effectiveness of the proposed approach in managing the traffic flow of a heterogeneous traffic system.