Innovative Topics and Future Challenges in Traffic Control

Wednesday, July 15, 11.00-13.00


Cecilia Pasquale, Simona Sacone, and Silvia Siri, University of Genova, Italy


The methods and algorithms that have been developed in the last decades for road traffic control are very relevant for their applicative implications for citizens and travelers, with immediate impacts on safety, quality of life, environment, use of energy resources, and costs. At the same time, such control methods are quite significant for the associated methodological challenges, since a road traffic system is an interesting application case (large-scale complex system with distributed state and bounded control) for all the researchers involved in control theory. While, on the one hand, a wide literature on traffic control has been developed over the past decades, the recent technological developments in sensors, processors, actuators and communication devices, as well as the new interest of companies and authorities for automated vehicle technologies, entail the update and development of new traffic control algorithms and architectures. Indeed, automated and connected vehicles will definitely change the features of the traffic networks in the next decades, and this will have a remarkable impact on traffic control strategies. If the future scenario in thirty years is expected to be a smart road environment with self-driving vehicles, the next future will surely present a mixed traffic in which traditional vehicles and intelligent vehicles will coexist. The design of control strategies for such a mixed scenario is one of the most urgent issues to be addressed by researchers in road traffic control. 


Opening of the Tutorial Session on “Innovative Topics and Future Challenges in Traffic Control”
Speakers: Cecilia Pasquale, Simona Sacone, and Silvia Siri, University of Genova, Italy
Wednesday July 15, 11:00-11:10

The main goal of this Tutorial Session is to provide an overview of the state-of-the-art in traffic control methods and to highlight the main research challenges due to the recent technological improvement in the automotive sector. These aspects will be discussed in six talks given by prominent and experienced researchers in the area of road traffic control.

Innovative topics and Future Challenges in Traffic Control: an Introduction
Speaker: Antonella Ferrara, University of Pavia, Italy
Wednesday, July 15, 11:10-11:25

The scientific, technological, social, and economic impact of successful research in road traffic control is very significant, with immediate effects on safety, quality of life, environment, use of energy resources, and transportation costs. Yet, the development of effective methods and algorithms for road traffic management has to face notable methodological challenges. In addition, the type of traffic control strategies developed so far, the “classical approaches”, need now to be updated and adapted to consider the fast development in automotive technologies, traffic sensors, data processing, and communications. This tutorial introductory presentation will address all these aspects, starting from an overview of classical traffic control concepts to arrive at encompassing emerging research trends at different scales: from the management of very large and complex traffic networks to the control of individual connected automated vehicles, also including heavy-duty vehicles, in order to have a beneficial influence on traffic dynamics. The major emerging topics in road traffic control, of which this presentation will provide an overview, will be discussed in more details in the subsequent presentations.

PDE Backstepping and Extremum Seeking for Stabilization and Optimization of Congested Traffic
Speaker: Miroslav Krstic, University of California, San Diego, CA, USA
Wednesday, July 15, 11:25-11:40

PDE backstepping has recently enabled the stabilization of congested upstream traffic, with ramp metering, for the state-of-the-art Aw-Rascle-Zhang PDE model. We present the ideas behind extending such control strategies to ARZ-like PDE models of multi-lane freeways and of freeways with multiple classes of vehicles (e.g., cars and trucks) or drivers (e.g., young and old, or human-driven and driverless). Combined with predictor feedback design, PDE backstepping is also applied to regulating a moving traffic shock where light traffic is upstream, heavy traffic is downstream, and a density discontinuity interface would propagate upstream in the absence of control. The interface position is governed by an ODE, with density states on both sides governed by the Lighthill-Whitham-Richards PDE. For traffic bottleneck problem caused by lane-drop, extremum seeking control, a non-model based approach for real-time optimization, is employed to regulate the incoming flow in the LWR model so that the outgoing traffic at the downstream bottleneck area is discharged at its maximum feasible steady flow rate.

Traffic Control for the Improvement of Sustainability in Freeway Networks: a Bibliometric Analysis
Speaker: Simona Sacone, University of Genova, Italy
Wednesday, July 15, 11:40-11:55

Sustainability is becoming a key factor in the design and control of traffic systems. In the field of traffic control for freeway networks, the performance criteria explicitly included in traffic control schemes to address sustainability are mainly referred to the reduction of the environmental impact and the decrease of accidents. These control objectives are generally considered together with the conventional goal of traffic control, i.e. the minimization of the total traveling times for drivers. In this talk, a bibliometric study based on keyword counting will be discussed, regarding the research works on freeway traffic control developed to improve sustainability. The main goal of such analysis is to evaluate how the interest of researchers on sustainability-related freeway traffic control has evolved over the last decades and which traffic control measures have been mostly used to address such issues.

Challenges in Control of Large-scale Freeway Networks
Bart De Schutter, Delft University of Technology, The Netherlands
Wednesday, July 15, 11:55-12:10

In this presentation, we discuss the main challenges in model-based control of large-scale freeway networks, which mostly pertain to a trade-off between control performance and computational effort. We consider several ways to address these challenges: we present a multi-level, multi-agent approach for model-based predictive control for large-scale networks, with a focus on freeway traffic networks. We also touch upon efficient macroscopic traffic models and various ways to speed up the MPC optimization such as approximations, distributed optimization, parametrized MPC, etc.

Traffic Control Algorithms for Mixed Vehicle Traffic
Speaker: Ioannis Papamichail, Technical University of Crete, Greece
Wednesday, July 15, 12:10-12:25

Connectivity is rapidly becoming a critical part of road transportation systems. Vehicle original equipment manufacturers (OEMs) are investing heavily in automation, while software is becoming a critical component of vehicles. At the same time, traffic and road authorities are seeking new technology solutions to increase safety and to reduce traffic congestion and emissions. These solutions are often dependent on vehicle functionality and the ability to provide various types of support for drivers and vehicles. As a result, there is a need to prepare the road infrastructure to support the coexistence of conventional and automated vehicles, targeting to the transition period when the penetration rate of connected automated vehicles (CAVs) will gradually increase.
A number of control strategies, targeting maximum throughput at motorway bottleneck locations, was developed by the DSSL group within the FP7 ERC project TRAMAN21. These include an Adaptive Cruise Control (ACC) parameter adaptation strategy, a Mainstream Traffic Flow Control (MTFC) strategy and a Lane-Change Advice (LCA) strategy. Preliminary investigations demonstrated that the proposed strategies improve the motorway traffic flow efficiency significantly, even for low penetration of CAVs. 
A library of software tools, that implement the control algorithms in a generic way for any network topology given by the user, has been developed recently within the H2020 project INFRAMIX. Extended simulations are currently performed using an advanced simulation environment (VSimRTI - V2X Simulation Runtime Infrastructure) developed by FOKUS. VSimRTI is a framework for the assessment of new solutions for cooperative intelligent transportation systems. The control algorithms are thoroughly assessed via comprehensive investigations involving a large variety of penetration rates for CAVs, infrastructure types and capabilities, and traffic conditions (free flow, critical, congested). Several of the developed tools will soon reach maturity allowing for their immediate practical usage, e.g. at the project’s test sites, as well as further exploitation.

Highway Traffic Control using Heavy-duty Vehicle Platooning
Speaker: Karl Henrik Johansson, KTH Royal Institute of Technology, Sweden
Wednesday, July 15,12:25-12:40

Traffic congestion has been a significant problem for a long time, with a wide array of negative effects on the society. Historically, there have been a number of classic traffic control methods, such as ramp metering and variable speed limits, that have been used to reduce traffic congestion with good effect. In recent years, the advent of autonomous vehicles and proliferation of V2X communication holds promise to revolutionize every part of the traffic control loop. Since automated heavy-duty vehicle platooning is currently in the process of being implemented and evaluated by several truck manufacturers world-wide, it is important to understand and analyze what effect this emerging technology will have on the rest of the traffic, and also how the rest of the traffic will affect platooning operations. In this talk, we describe and model the mutual influence that heavy-duty vehicle platoons and the rest of the traffic have on each other, and explore what effects ubiquitous platooning might have. We may expect that introduction of additional heavy-duty vehicles will have an adverse effect on the traffic. However, the fact these vehicles will be automated and communicating allows driving with low intervehicular gaps, reducing the space they would take on the road. Furthermore, since these vehicles could be controlled directly from the infrastructure, we would be able to potentially use them as actuators, enabling the introduction of new traffic control methods.

Interactive Live Session
Wednesday, July 15, 12:40-13:00