WP3: Motorway Traffic Control

Among the set of VACS that have a direct impact on motorway traffic flow (see WP1 for a wide classification), some functionalities can be exploited for the development of novel control strategies, which may generate improvements to motorway traffic conditions.

Task 3.1: Hierarchical Control Structure

3.1.1 Hierarchical Control Design: The coordinated and integrated control of motorway traffic, exploiting various VACS features and options, is tackled by defining a hierarchical control framework. It is assumed that VACS provide the possibility to obtain more accurate traffic state estimates and that they are able to perform advanced control tasks, according to the centrally-computed Variable Speed Limits (VSLs) and "change-lane" orders or recommendations. The structure is composed by different layers: the Adaptation and prediction layer processes data retrieved from the motorway and provides information for the lower layers; the Optimisation layer includes an optimisation problem that takes into account different control actions, and is employed according to a Model Predictive Control scheme, in order to reject unexpected external disturbances and modelling errors; the Application layer converts the output of the Optimisation layer to corresponding tasks assigned to specific actuators on the road or in the equipped vehicles. Further decentralised Local-layer controllers can be added, characterised by a shorter control period, in order to faster react to external disturbances. Demonstration examples are built with the aid of a microscopic traffic simulator, to show the improvements that can be achieved by the application of the proposed hierarchical control strategy.

  • Roncoli, C., Papamichail, I., Papageorgiou, M.: Model predictive control for multi-lane motorways in presence of VACS, Proceedings of the 17th International IEEE Conference on Intelligent Transportation Systems (ITSC 2014), Qingdao, China, 8-11 October 2014, pp. 501-507. -LINK-
  • Roncoli, C., Papamichail, I., Papageorgiou, M.: Model predictive control for motorways in presence of mixed manual and VACS-equipped vehicles. 18th meeting of the EURO Working Group on Transportation (EWGT2015), Delft, The Netherlands, 14-16 July 2015, Transportation Research Procedia, vol. 10, 2015, pp. 452-461.
  • Roncoli, C., Papamichail, I., Papageorgiou, M.: Hierarchical model predictive control for multi-lane motorways in presence of Vehicle Automation and Communication Systems. Transportation Research Part C, 2016, vol. 62, pp. 117-132.

3.1.2 Traffic State Estimation: The availability of reliable real-time measurements or estimates of the traffic state is a prerequisite for successful traffic control. In conventional traffic, the necessary measurements are provided by spot sensors (based on a variety of possible technologies), which are placed at specific motorway locations. With the introduction of VACS of various kinds, an increasing number of vehicles become "connected", i.e. enabled to send (and receive) real-time information to (and from) a local or central monitoring (and control) centre. Thus, connected vehicles may communicate their position, speed and other relevant information, i.e. they can act as mobile sensors. This allows for a sensible reduction (and, potentially, elimination) of the necessary number of spot sensors, which would lead to sensible reduction of the purchase, installation and maintenance cost for traffic monitoring; while, at the same time, improving the estimation quality. The present TRAMAN21 activity concerns the development of reliable and robust traffic flow estimation methods and tools, which exploit information provided by connected vehicles and reduce the need for spot sensor measurements under all penetration rates of connected vehicles.

A macroscopic model-based approach for estimation of the traffic state, specifically of the (total) density and flow of vehicles, is developed for the case of "mixed" traffic, i.e., traffic comprising both ordinary and connected vehicles. The development relies on the following realistic assumptions: (i) The density and flow of connected vehicles are readily known at the traffic control centre (TCC) on the basis of their regularly reported positions; and (ii) the average speed of conventional vehicles is roughly equal to the average speed of connected vehicles; and is hence known, thanks to the speed reports received regularly from connected vehicles. Thus, complete traffic state estimation (for arbitrarily selected segments in the network) may be achieved by merely estimating the percentage of connected vehicles with respect to the total number of vehicles. A model is derived, which describes the dynamics of the percentage of connected vehicles, utilizing well-known conservation law equations that describe the dynamics of the density of connected vehicles and of the total density of all vehicles. Based on this model, which is a linear time-varying system, an estimation algorithm for the percentage of connected vehicles is developed employing a Kalman filter. The estimation methodology is validated through simulation using a second-order macroscopic traffic flow model as ground truth for the traffic state. The approach calls for a minimum of spot sensor-based total flow measurements according to a variety of possible location configurations.

The developed traffic state estimation methodology is evaluated and tested in two case studies that are built using data stemming from different real experiments. The first case, based on real microscopic traffic data collected within the Next Generation SIMulation (NGSIM), aims at verifying the capability to estimate the traffic state at high space-resolution (i.e., considering segments in the order of 50 m in length), using aggregate information retrieved from moving vehicles, that is, the speeds of connected vehicles and the total flow measured at the entrance and exit of the highway stretch; the proposed estimation scheme is tested for different penetration rates of connected vehicles. In the second case study, a longer highway stretch and a longer time horizon are considered, using information coming from fixed detectors, namely speed measurements at every segment (emulating connected vehicle reports) and a small amount of flow measurements. This permits to verify the capability of detecting the onset of a congestion created within the network, which is one of the most crucial and challenging aspects in case this approach is exploited in conjunction with a traffic control strategy. In both case studies, the performances are satisfactory and the obtained results demonstrate the effectiveness of the methodology, both in qualitative and quantitative terms.

In addition, an alternative estimation scheme has been developed for direct estimation of the total traffic density in highways utilizing only average speed measurements reported by connected vehicles and a minimum number (sufficient to guarantee observability) of spot sensor-based total flow measurements, thus relaxing the requirement of the percentage estimator of measuring flows and densities for connected vehicles. This, alternative, approach exploits the fact that the dynamics of the total density, as described by the conservation law equation, can be described by a linear time-varying system with known parameters on the basis of the assumption that the average speed of conventional vehicles is roughly equal to the average speed of connected vehicles, and, consequently, it can be obtained at the (local or central) traffic monitoring and control unit from connected vehicles' reports. Based on this linear time-varying system, a Kalman filter is employed for the estimation of the total density. It is further demonstrated that the developed, alternative approach allows a variety of different measurement configurations, by showing that traffic state estimation is achieved when additional mainstream total flow measurements are employed to replace a corresponding number of total flow measurements at on-ramps or off-ramps. The performance of the estimation scheme is validated through simulations using a well-known second-order traffic flow model as ground truth for the traffic state, as well as using NGSIM real microscopic traffic data.

Currently, both developed traffic estimation methodologies are evaluated with a much more detailed microscopic simulation platform, thus considering a more realistic representation of all involved real-time measurements. Several different realistic scenarios are employed and evaluated, including: i) the case where the information needed for estimation is retrieved only from speeds and positions reported at random time instances from connected vehicles, ii) varying penetration rates of connected vehicles, iii) cases where for some time intervals there is no information coming from connected vehicles, iv) the case of connected vehicles equipped with an Adaptive Cruise Control (ACC) feature, v) the performance comparison between the two estimation schemes (utilizing, e.g., the relative estimation error as a quantitative measurement of performance), and vi) the study of the effect in estimation of model parameters variation.

  • Bekiaris-Liberis, N., Roncoli, C., Papageorgiou, M.: Highway Traffic State Estimation With Mixed Connected and Conventional Vehicles Using Speed Measurements. 18th International IEEE Conference on Intelligent Transportation Systems (ITSC 2015), Las Palmas de Gran Canaria, Spain, 15-18 September 2015, pp. 2806-2811.
  • Roncoli, C., Bekiaris-Liberis, N., Papageorgiou, M.: Motorway traffic state estimation using speed measurements: Case studies on NGSIM data and motorway A20 in the Netherlands. Proceedings of the 95th Annual Meeting of the Transportation Research Board (TRB), Washington, D.C., USA, 10-14 January 2016, paper no. 16-2071.
  • Bekiaris-Liberis, N., Roncoli, C., Papageorgiou, M.: Highway traffic state estimation with mixed connected and conventional vehicles. IEEE Transactions on Intelligent Transportation Systems, accepted.
  • Roncoli, C., Bekiaris-Liberis, N., Papageorgiou, M.: Use of speed measurements for highway traffic state estimation: Case studies on NGSIM data and highway A20, Netherlands. Transportation Research Record, 2016, vol. 2559, pp. 90-100.
  • Bekiaris-Liberis, N., Roncoli, C., Papageorgiou, M.: Highway Traffic State Estimation with Mixed Connected and Conventional Vehicles. 14th IFAC Symposium on Control in Transportation Systems, Istanbul, Turkey, 18-20 May 2016, pp. 309-314.
  • Fountoulakis, M., Bekiaris-Liberis, N., Roncoli, C., Papamichail, I., Papageorgiou, M.: Highway traffic state estimation with mixed connected and conventional vehicles: Microscopic simulation-based testing. 2016 IEEE 19th International Conference on Intelligent Transportation Systems, Rio de Janeiro, Brazil, November 1-4, 2016, pp. 1761-1766.

Task 3.2: Local Control Level

3.2.1 Mainstream Traffic Flow Control:The use of Mainstream Traffic Flow Control (MTFC) with Variable Speed Limits (VSL) has been investigated in the last decade as a potential control strategy to mitigate traffic congestion. Some works carried on within this Task involved the review and application of an optimal control approach and a feedback control approach, both proving capable of improving significantly the traffic conditions of a motorway stretch in a simulation environment. Also microscopic simulation has been used, permitting to discover important aspects that had not been previously captured using macroscopic simulation. Another studied aspect is the integration of local Ramp Metering (RM) and MTFC, via the extension of an existing local cascade feedback controller for MTFC-VSL by use of a split-range-like scheme that allows different control periods for RM and MTFC-VSL.

  • Müller, E. R., Carlson, R. C., Kraus, W., Jr., Papageorgiou, M.: Microscopic simulation analysis of mainstream traffic flow control with variable speed limits. Proceedings of the 16th International IEEE Conference on Intelligent Transportation Systems (ITSC 2013), The Hague, The Netherlands, October 6-9, 2013, pp. 998-1003. -LINK-
  • Carlson, R. C., Papamichail, I., Papageorgiou, M.: Mainstream traffic flow control on freeways using variable speed limits. Transportes (Rio de Janeiro), 2013, vol. 21, n. 3, pp. 56-65. -LINK-
  • Carlson, R. C., Papamichail, I., Papageorgiou, M.: Integrated feedback ramp metering and mainstream traffic flow control on motorways using variable speed limits. Transportation Research Part C, 2014, vol. 46, pp. 209–221. -LINK-
  • Wang, Y., Kan, Y, Papageorgiou, M., Papamichail, I.: Local ramp metering with distant downstream bottlenecks: a comparative study, Proceedings of the 17th International IEEE Conference on Intelligent Transportation Systems (ITSC 2014), Qingdao, China, 8-11 October 2014, pp. 768-773. -LINK-
  • Müller, E. R., Carlson, R. C., Kraus, W., Jr., Papageorgiou, M.: Microsimulation analysis of practical aspects of traffic control with variable speed limits. IEEE Transactions on Intelligent Transportation Systems, 2015, vol. 16, pp 512-523. -LINK-
  • Kan, Y., Wang, Y., Papageorgiou, M., Papamichail I.: Local ramp metering with distant downstream bottlenecks: A comparative study. Transportation Research Part C, 2016, vol. 62, pp. 149-170.

3.2.2 Cooperative Merging: A cooperative merging system was developed, while its performance and its impact on highway capacity were evaluated. Two algorithms have been developed and evaluated for the cooperative merging of vehicles in highways, within the framework of microscopic traffic simulator Aimsun; they were initially designed and tested for a simple network, consisting of a single mainstream lane and a single on-ramp, leading to an acceleration lane. The proposed algorithms can simulate vehicle-to-infrastructure (V2I) and vehicle-to-vehicle (V2V) communication, allowing for the cooperation of all vehicles traveling on the main motorway and those coming from the on-ramp. Both algorithms were tested for a variety of on-ramp flows and desired time-gaps. The simulation results showed that the two algorithms have very similar performance, although different gaps may be selected by the entering cars, according to the algorithm used. However, the 2nd algorithm is closer to real merging, since it evaluates also forward gaps and mitigates unnecessary decelerations. For all the cases considered, the merging was very smooth, without unreasonable and irrational situations.

A further extension of the previous methodology was considered for the application of cooperative merging principle to the lane changing procedure. To this end, a cooperative lane changing system was developed, which utilizes modified versions of the aforementioned cooperative merging algorithms, to establish a cooperative lane changing procedure at a region before an on-ramp. The goal is to decrease the flow in the right lane, render the (cooperative) merging of the incoming vehicles more efficient, and produce a uniform distribution of the flow between the lanes downstream the on-ramp. The proposed methodology was evaluated within the framework of microscopic traffic simulator Aimsun, with very promising results.

A second step was to apply optimal control theory to the problem of cooperative merging. Assuming that a single vehicle is appropriately equipped and has access to data regarding the movement of neighbouring vehicles (via sensors or communication), and that it knows its putative leader, analytical solutions are developed for the optimal trajectory of the vehicle, from the time it enters the on-ramp until it successfully merges behind its putative leader. Optimality is defined in a multi-objective setting on different levels, which includes driver comfort, acceleration/deceleration effort and fuel consumption. The cost function to be minimized includes the weighted sum of the squares of the acceleration, jerk and derivative of the jerk. The tunable weights, reflecting the desired importance of each sub-criterion, may serve as parameters to match the individual driver’s preferences. The proposed methodology was first developed for a pair of cooperating vehicles, a merging one and its putative leader. Moreover, an alternative solution procedure via a time-variant Linear-Quadratic Regulator approach was also developed. A Model Predictive Control (MPC) scheme was utilized to compensate possible disturbances in the trajectories of the cooperating vehicles, whereby the analytical optimal solution is applied repeatedly in real time, using updated measurements, until the merging procedure is actually finalized. The proposed MPC scheme for a pair of vehicles was finally extended and applied to a set of vehicles entering the cooperation area, based on a prescribed merging sequence. The extended methodology includes the use of an ACC controller for all vehicles, while the MPC one is used along with the ACC inside the cooperation area.

This concept can be extended to the case of many vehicles, where, apart from the individual vehicle trajectories, the optimal merging sequence can be also found. In order to compute an optimal merging sequence, different combinations of merging and mainstream vehicles should be evaluated, within the context of optimal trajectories, as previously presented. However, the estimated arrival times to the merging point may cause several conflicts between the vehicles. To this end, groups of successive vehicles inside the cooperation area are formed, with the vehicles in the same group interacting to each other due to a conflict. Instead of checking the interactions of all possible pairs of vehicles inside the cooperation area, only the vehicles inside each one of the formed groups are considered (in order to decrease the computational time). The arrival times of the vehicles belonging to the same group are subsequently modified, in order to resolve the conflict. These times are then used to compute the optimal trajectories of all vehicles in the same group from their current position to the merging point. For all the vehicles in the same group, all the alternative trajectories are computed and their cost for the corresponding trajectory is evaluated. The sequence of vehicles in the same group with the lowest total cost is selected for the corresponding time iteration.

  • Ntousakis, I. A., Porfyri, K., Nikolos, I. K., Papageorgiou, M.: Assessing the impact of a cooperative merging system on highway traffic using a microscopic flow simulator, Proceedings of the ASME 2014 International Mechanical Engineering Congress and Exposition (IMECE2014), Montreal, Quebec, Canada, 14-20 November 2014, . Paper No. IMECE2014-39850. -LINK-
  • Ntousakis, I. A., Nikolos, I. K., Papageorgiou, M.: Optimal vehicle trajectory planning in the context of cooperative merging on highways. Transportation Research Part C, 2016, vol. 71, pp. 464-488.
  • Ntousakis, I.A., Nikolos, I.K., Papageorgiou, M.: Cooperative vehicle merging on highways – model predictive control. Proceedings of the 96th Annual Meeting of the Transportation Research Board (TRB), Washington, D.C., USA, 8-12 January 2016, paper no. 17-00930.

3.2.3 Development of Advanced Nonlinear Feedback Control Concepts: This stream of work is aimed to exploit and apply recent nonlinear feedback control concepts to a variety of traffic control problems. The following topics have been studied up to now: robust global stability of the uncongested equilibrium point of uncertain traffic networks with no restriction on the topology of the network; algorithms for the verification of robust global stability of the uncongested equilibrium point for open-loop freeway models; construction of nonlinear model-based, robust (global or local) feedback stabilizers for the uncongested equilibrium point of (a) simple traffic models and (b) uncertain freeway models which follow the principles of the Cell Transmission Model; construction of Control Lyapunov Functions for uncertain freeway models and utilization of Lyapunov functions for the study of robustness of the closed-loop system with respect to external inputs; construction of nonlinear global feedforward feedback stabilizers which guarantee the complete disturbance rejection of constant external inputs for freeway models; comparison of the achieved performance of the newly proposed nonlinear feedback stabilizers and existing feedback control methodologies for freeway models. Interconnections of freeways are also studied under very weak assumptions leading to highly nonlinear traffic networks. The study of interconnections of freeways is performed by using single or vector Lyapunov functions for the overall system and the design of robust nonlinear feedback stabilizers for the uncongested equilibrium point can be achieved under minimal requirements for the topology of the network. The obtained results allow for the study of the following research topics in the near future: the construction of adaptive feedback laws for the robust stabilization of the uncongested equilibrium point of uncertain traffic networks; the construction of nonlinear observers which can be used in dynamic feedback stabilizers for the robust stabilization of the uncongested equilibrium point of uncertain traffic networks; the utilization of inverse optimality arguments for the construction of optimal-performance feedback stabilizers for freeway models. Extensive simulation studies are performed for the testing of the performance of all proposed nonlinear feedback stabilizers. The simulation studies are not limited to uncertain first order freeway models (the case covered by theoretical results) but also use second-order models (e.g. METANET).

  • Karafyllis, I., Papageorgiou, M.: Stability results for simple traffic models under PI-regulator control. IMA Journal of Mathematical Control and Information, 2016, vol. 33(2), pp. 209-229.
  • Karafyllis, I., Papageorgiou, M.: Stability investigation for simple PI-controlled traffic systems. Proceedings of 6th International Symposium on Communications, Controls, and Signal Processing, Athens, Greece, May 21-23, 2014, pp. 186-189. -LINK-
  • Karafyllis, I., Papageorgiou, M.: Global Exponential Stability for Discrete-Time Networks with Applications to Traffic Networks. IEEE Transactions on Control of Network Systems, 2015, vol. 2(1), pp. 68-77. -LINK-
  • Karafyllis, I., Kontorinaki, M. and Papageorgiou, M.: Global exponential stabilization of freeway models, International Journal of Robust and Nonlinear Control, 2016, vol. 26, pp. 1184-1210.
  • Karafyllis, I., Kontorinaki, M. and Papageorgiou, M., "Global exponential stabilization of freeway models", Proceedings of the European Control Conference 2015, Linz, Austria, pp. 2645-2650.
  • Karafyllis, I., Kontorinaki, M. and Papageorgiou, M., "Robust Global Adaptive Exponential Stabilization of Discrete-Time Systems with Application to Freeway Traffic Control", submitted.
  • Karafyllis, I., Kontorinaki, M., Papageorgiou, M.: Nonlinear adaptive control scheme for discrete-time systems with application to freeway traffic flow networks. 15th Annual European Control Conference (ECC 2016), Aalborg, Denmark, June 29 - July 1, 2016, pp. 275-282.
  • Kontorinaki, M., Karafyllis, I., Papageorgiou, M.: Nonlinear feedback control for general acyclic traffic networks.
  • Kontorinaki, M., Karafyllis, I., Papageorgiou, M.: Global exponential stabilization of acyclic traffic networks.

Task 3.3: Network-wide Traffic Control

The activities within this task involve the development of an integrated traffic control problem, addressed through the formulation of a linearly-constrained convex Quadratic Programming problem, based on a novel first-order multi-lane model for motorways that has been appropriately designed (see Activity 2.2.2). Within this problem, the use of some VACS permits the implementation of an increased range of control tasks. Specifically, VACS can be used to improve Ramp Metering and Mainstream Traffic Flow Control via Variable Speed Limits; and to perform Lane Changing Control. The use of VACS may drastically upgrade the possibilities of applying VSL, since each equipped vehicle can receive an individual speed limit that should be respected while driving in the current location, replacing the use of Variable Message Signs located on gantries. The developed integrated optimization problem is shown, in a number of different infrastructures and control settings, to exploit efficiently the synergy offered by the three included actuators towards achieving highly efficient control results.

  • Roncoli, C., Papageorgiou, M., Papamichail, I.: Motorway flow optimization in presence of vehicle automation and communication systems. Proceedings of the 1st International Conference on Engineering and Applied Science Optimization (OPT-i), Kos, Greece, June 4-6, 2014, pp. 519-529. -LINK-
  • Roncoli, C., Papageorgiou, M., Papamichail, I.: Optimal control for multi-lane motorways in presence of Vehicle Automation and Communication Systems. 19th IFAC World Congress (IFAC'14), Cape Town, South Africa, August 24-29, 2014, pp. 4178-4183. -LINK-
  • Roncoli, C., Papageorgiou, M., Papamichail, I.: Traffic flow optimisation in presence of vehicle automation and communication systems - Part II: Optimal control for multi-lane motorways. Transportation Research Part C, 2015, vol. 57, pp. 260-275.

Task 3.4: Large-Scale Virtual Test Site Demonstration

A large-scale real European motorway infrastructure is used as a virtual test site for simulation-based testing and demonstration of various modelling and control concepts that are being developed within TRAMAN21 for motorway traffic flow in presence of VACS. The aims of the related work are twofold: (a) To provide a realistic test bed for the suitability of various developed concepts, control strategies and their characteristics; (b) To demonstrate to the relevant research communities, road authorities and further stakeholders the achievable benefits of VACS and of the developed control approaches. The investigations make extensive use of real traffic data, including external demands (at mainstream and on-ramp origins) and detailed motorway data (lane-based flows, occupancies and speeds) for maximizing the realism and relevance of the obtained results.

3.4.1 Site Selection: Among various potential options, it was decided to use as a virtual testbed a stretch of the motorway A20 (from Rotterdam to Gouda) in The Netherlands. This infrastructure incorporates a non-trivial combination of lane-drops, on-ramps, and off-ramps, and is characterized by a variability of congestion patterns, including recurrent overload congestion during peak hours, occasional congestion due to over-spilling of off-ramp queues, as well as shock waves mounting from the downstream boundary. The infrastructure is equipped with closely-spaced lane-specific mainstream detectors (providing lane-based flows, occupancies and speeds at distances of 300 m on average), as well as flow detectors at on- and off-ramps. These characteristics reflect a challenging and well equipped test-bed and were the main reasons for the selection made. Appropriate and sufficient (multiple days) traffic data were made available for the purposes of TRAMAN21 by Prof. Bart van Arem (Technical University of Delft), which is gratefully acknowledged.

3.4.2 Microscopic Model Validation: In order to enable the usage of the motorway infrastructure for comparative evaluation of various TRAMAN21 tools, a comprehensive calibration exercise was carried out for a version of the microscopic simulator Aimsun, using the available lane-specific traffic data from the site. Specifically, to avoid some known shortcomings of microscopic simulators in complex merging situations, the simulator’s default models for car-following and lane-changing were replaced or appropriately modified; this was followed by careful and thorough model parameter estimation, based on both rigorous optimization tools and manual tuning. The calibration and validation procedure resulted in a set of parameters which match pretty accurately the lane-based traffic conditions for several days of available data and provides a realistic reference case for various TRAMAN21 Activities.

3.4.3 Traffic Estimation: The motorway infrastructure was used as one of the test and demonstration sites for the traffic state estimation schemes based on connected vehicles reports, that are developed within Activity 3.1.2. To this end, the available measured speeds were appropriately perturbed with noise to emulate connected vehicle reports. Moreover, a minimum number of required flow measurements were also used for producing comprehensive traffic information w.r.t. traffic densities and flow; which were compared to the ground truth of available real data. The obtained results confirmed the applicability and good performance of the estimation scheme.

3.4.4 Testing of Network-wide Traffic Control with VACS:The motorway infrastructure was used as the first test and demonstration site for the link-based integrated traffic control method developed within Task 3.3. The obtained results for this complex infrastructure were excellent and opened the way for farther enhancements and applications.

3.4.5 Development and Evaluation of Vehicle-Level Control Approaches:Various versions and aspects of vehicle-level control (Task 3.5), aiming, beyond fast individual vehicle advancement, also at benefitting the traffic flow efficiency, are ongoing on the basis of the motorway infrastructure, and, in particular, the validated microscopic model of Activity 3.4.2. This allows for insertion of an arbitrary number or percentage of vehicles with special behaviour (e.g. automated car-following or lane-changing behaviour) and development and evaluation of the impact of various interventions, such as real-time adaptation of vehicle-control parameters, including various trade-offs while optimizing individual vehicle versus total flow efficiency.

The following publications (related to all above Activities) use the selected motorway infrastructure as a test-bed:

  • Roncoli, C., Papageorgiou, M., Papamichail, I.: Motorway flow optimization in presence of vehicle automation and communication systems. Proceedings of the 1st International Conference on Engineering and Applied Science Optimization (OPT-i), Kos, Greece, June 4-6, 2014, pp. 519-529. -LINK-
  • Roncoli, C., Papageorgiou, M., Papamichail, I.: Optimal control for multi-lane motorways in presence of Vehicle Automation and Communication Systems. 19th IFAC World Congress (IFAC'14), Cape Town, South Africa, August 24-29, 2014, pp. 4178-4183. -LINK-
  • Roncoli C., Papageorgiou, M., Papamichail, I.: Motorway Traffic Flow Optimisation in Presence of Vehicle Automation and Communication Systems, In Engineering and Applied Sciences Optimization, Lagaros, N.D., Papadrakakis, M., (Eds.), Series Computational Methods in Applied Sciences, Springer International Publishing, 2015, Vol. 38, pp. 1-16.
  • Roncoli, C., Bekiaris-Liberis, N., Papageorgiou, M.: Motorway traffic state estimation using speed measurements: Case studies on NGSIM data and motorway A20 in the Netherlands. Proceedings of the 95th Annual Meeting of the Transportation Research Board (TRB), Washington, D.C., USA, 10-14 January 2016, paper no. 16-2071. -LINK-
  • Roncoli, C., Bekiaris-Liberis, N., Papageorgiou, M.: Motorway traffic state estimation using speed measurements: Case studies on NGSIM data and motorway A20 in the Netherlands. Transportation Research Record, accepted.
  • Perraki, G., Roncoli, C., Papamichail, I., Papageorgiou, M.: Evaluation of an MPC strategy for motorway traffic comprising connected and automated vehicles, submitted.

Task 3.5: Vehicle-level Control for Improved Traffic Flow

3.5.1 ACC-Vehicle Control: While developing the advanced capabilities of VACS, the automotive industry has mainly focused on designing algorithms that optimize the behavior of an individual vehicle, rather than on optimizing the overall behavior of the traffic network. This vehicle-centric approach may have little effect on the traffic flow efficiency in mild traffic conditions; however, it may degrade substantially the nominal traffic infrastructure capacity in heavy traffic conditions, as, for example, during rush hours, at bottlenecks locations, or due to unpredictable factors, such as accidents. Therefore, there is an urgent need to develop appropriate novel traffic management concepts which will enable the exploitation of existing or emerging VACS towards increased traffic flow efficiency.

We are currently investigating the development of advanced decision and control algorithms that enable for the actions of individual vehicle systems to lead, beyond the fulfillment of their individual goals, to the emerging optimization of the performance of the overall traffic system. The following two control approaches are adopted:

  1. Completely distributed: individual vehicles act based on their own decisions (without intervention of a superior coordinating unit); this includes two subcases: (a) presence of V2V communication; and (b) completely autonomous vehicles.
  2. Distributed with hierarchical information exchange: the traffic infrastructure may receive or transmit information via V2I communication, which is used by individual vehicles while making decisions.

In addition, we are currently developing a predictor-based ACC design for compensation of arbitrarily large actuator and sensor delays in vehicular systems utilizing measurements of the relative spacing as well as of the speed and acceleration of each individual vehicular system. The importance of addressing this problem becomes clear when one considers that such delays are ubiquitous in traffic systems and may be due to engine response, throttle or brake actuator, computational time, radar or lidar systems, wheel speed sensors, and sampling of measurements. The predictor-based ACC law guarantees string stability of platoons of vehicular systems for any delay value. This is crucial since it implies that even in the presence of such delays the traffic capacity and traffic throughput do not decrease, comfort and safety do not deteriorate, and fuel consumption does not increase.

  • Bekiaris-Liberis, N., Roncoli, C., Papageorgiou, M., Predictor-based adaptive cruise control design. 2016 IEEE Multi-Conference on Systems and Control, Buenos Aires, Argentina, September 19-22, 2016, pp. 97-102.
  • Bekiaris-Liberis, N., Roncoli, C., Papageorgiou, M., Predictor-based adaptive cruise control design with integral action.

3.5.2 Automated Vehicle Trajectory Optimisation: This part of the research work aims at the optimization of the (individual or collective) trajectories of automated vehicles in a way that accounts for their implications for the emerging traffic flow efficiency. With the advent of highly automated vehicles, the trajectory optimization is a significant aspect, since it directly affects traffic flow characteristics and safety as well as the driver acceptance or not of the automated mode of driving. The scope of this work is to develop trajectory generation strategies that fulfill the goals of automated driving while, at the same time, exploiting the presence of VACS-equipped vehicles to improve traffic flow efficiency. Special care is devoted to real-time feasibility concerns. The plan for this work is structured to include various alternative control strategies and computational architectures as well as different traffic scenarios and penetration rates. Specifically, the cases of individual vehicle control, control of clusters of vehicles and hierarchical control are some of the envisaged options. A range of different highway movement scenarios are being investigated, such as merging and lane-changing of vehicles, lane-drops and other types of bottlenecks.

3.5.3 Feedback-based lane assignment: Throughput maximization at bottleneck areas is crucial due to its potential for alleviating or avoiding traffic congestion. Since connected vehicles may be asked (or recommended) to perform lane changing, it is natural to look into possibilities of exploiting such capabilities for throughput maximization at bottleneck areas via optimal vehicle lane assignment. Feedback control strategies may be employed to guarantee robust regulation of the traffic flow at optimal lane-assignment set-points at bottleneck locations.

We are currently developing a feedback control strategy for throughput maximization at bottleneck areas via manipulating the lane changing behavior of connected or automated vehicles, such that the density at specific bottleneck areas is regulated at given optimal set-points. The control strategy is comprised of a feedback and a feedforward part. The feedback part is designed using a standard Linear Quadratic Regulator. The feedforward part, namely the desired set-points for the density at specific positions, is generated via an Extremum Seeking algorithm.

  • Roncoli, C., Bekiaris-Liberis, N., Papageorgiou, M.: Optimal lane-changing control at motorway bottlenecks. 2016 IEEE 19th International Conference on Intelligent Transportation Systems, Rio de Janeiro, Brazil, November 1-4, 2016, pp. 1786-1791.
  • Roncoli, C., Bekiaris-Liberis, N., Papageorgiou, M.: Lane-changing feedback control for efficient lane assignment at motorway bottlenecks. submitted.

Cooperation with visitors

MPC-based eco-driving model for vehicles

In cooperation with PhD Candidate Weiming Zhao

Eco-driving model can be applied to vehicles in various circumstances, e.g. while merging in a motorway or approaching an urban intersection, and may be facilitated via V2V or V2I communication and vehicle automation functions. For the optimal trajectory planning of a single vehicle, appropriate optimal control problems for different eco-driving models have been formulated. The problems are solved by the recourse of different numerical methods, among which the feasible direction method proved clearly as the most efficient. For the trajectory planning of a platoon of vehicles in urban roads, a model predictive control model was developed, which utilizes the traffic signal information and minimizes fuel consumption for the whole platoon.

Real-time Traffic Control for Reversible Motorway Lanes

In cooperation with PhD Candidate José Ramón D. Frejo

The principle of reversible lanes operation is to match available capacity to the traffic demand, taking advantage of the unused capacity in the minor-flow direction lanes to increase the capacity in the major-flow direction. Most reversible lane systems have been developed and managed based primarily on experience, professional judgment, and empirical observation. In this work, an extension of the second-order traffic flow model METANET to model reversible lanes on freeways was first undertaken. Based on this model, two kinds of dynamic controllers were developed. The first one is an easy-to-implement logic-based controller which takes into account the congestion lengths generated by the reversible lane bottleneck and uses this information for the real-time operation of the lanes. The second one is a discrete Model Predictive Control (MPC) which minimizes the Total Time Spent (TTS) of the modelled network subject to constraints for the maximum generated bottleneck queues. The proposed model and control algorithms were simulated and tested using loop detector data collected over a section of the SE-30 freeway in Seville, Spain. The modelled network includes the Centenario Bridge, which is a bottleneck with a reversible lane that creates recurrent congestion during the morning rush-hour period. The results show that both proposed controllers (which can be computed in a short time) substantially reduce this congestion.

  • Domínguez Frejo, J. R., Papamichail, I., Papageorgiou, M., Camacho, E.: Macroscopic modeling and control of reversible lanes on freeways. IEEE Transactions on Intelligent Transportation Systems, 2015, vol. 17, pp. 948-959.
Optimal control of multi-class traffic flow models

In cooperation with PhD Candidate Cecilia Pasquale

The regulation of freeway traffic by means of optimal control techniques is considered for a model that includes two classes of vehicles (cars and trucks). A first innovative aspect of the proposed approach is the adopted objective function, in which, besides the reduction of traffic congestion (which is typically considered in classic traffic control schemes), the minimisation of traffic emissions is also included. A multi-objective optimal control problem is therefore defined, assuming that specific control actions (ramp metering) are taken for each vehicle class; an appropriate feasible direction algorithm is applied, testing then the effectiveness of the proposed approach by means of macroscopic simulations.

  • Pasquale, C., Papamichail, I., Roncoli, C., Sacone, S., Siri, S., Papageorgiou, M.: A nonlinear optimal control approach for two-class freeway traffic regulation to reduce congestion and emissions. 14th annual European Control Conference (ECC 2015), Linz, Austria, 15-17 July 2015, pp. 2651-2656.
  • Pasquale, C., Papamichail, I., Roncoli, C., Sacone, S., Siri, S., Papageorgiou, M.: Two-class freeway traffic regulation to reduce congestion and emissions via nonlinear optimal control. Transportation Research Part C, 2015, vol. 55, pp. 85-99.
Dynamic Estimation of O-D Flows from Traffic Counts

In cooperation with Assistant Professor Vittorio Marzano

Dynamic estimation of origin-destination streams from available link traffic counts remains, despite extended past research work, a significant problem for various uses within Transportation. In this work, a number of potential novel approaches were discussed and elaborated to tackle the problem, including new ways of applying the Extended Kalman Filter, Moving-Horizon Estimation and Constrained Optimisation. Corresponding developments and tests for an Italian motorway are in progress.

Last updated on 13/02/2017.