Autori
Alessio Iovine, Elena De Santis and Maria Domenica Di Benedetto
Abstract
Traffic control is one of the most studied problems in engineering because of its impact on human life: progress in the knowledge and control of traffic systems would improve life quality, especially in an urban environment. The goal of traffic control is to manage the flow of cars in highways so that a number of quantities such as congestion, emissions, travel time reduction and safety are traded-off in an optimal fashion. Driver support systems such as Adaptive Cruise Control (ACC) systems, and Advanced Driver-Assistance Systems (ADAS) provide full or partial driver assistance control traffic. Once ADAS and ACC systems are introduced in cars, the behavior of the cars should follow normal traffic dynamics in order to be able to interact with human drivers. To do so, we believe driver support systems should mimic driver behavior.
The objective of this project is developing an ACC model able to mimic the human driver behavior with respect to comfort while ensuring a proper safety level. With this intent, classical models will be used. Over the years, a multitude of traffic control systems have been proposed. Those models can be classified on the basis of their level of abstraction (macroscopic, microscopic or mesoscopic) or the adopted control strategy (centralized or decentralized) and have been utilized with different methodologies for managing a large variety of problems.
We first analyze the properties of the overall hybrid system when applying state feedback control laws, using only local information. Afterwards, we define a mesoscopic model, where the control action depends not only on individual information but also on the traffic flow, which is a macroscopic quantity. The macroscopic information can either be provided by a centralized traffic supervisor, or it can be gathered, elaborated and transmitted by the vehicles themselves, when they are connected and can exchange information. We consider the second case.
