As a part of the global plan to achieve environmental sustainability, eco-driving defines a set of driving behaviors that benefit both the drivers, by reducing fuel or energy consumption, and the environment, by mitigating emissions. This paper refers to urban eco-driving concepts for electric vehicles with regenerative braking, aiming to mitigate energy consumption on trajectories from their current state to a fixed final state. Such trajectories are determined as solutions of an optimal control problem having the acceleration of the electric vehicle as control input, and the position and speed as state variables. A power-based energy model is used to calculate the energy consumption and the energy regeneration while braking, resulting in a non-smooth model with some discontinuities. In order to solve the problem, different approaches are considered, based on different approximations of the power-based model. Such approaches are then extended to address the stop point problem, considering a traffic signal placed on the way of the vehicle. A comparative analysis is reported and discussed in the paper for different initial state and signal timing scenarios. In particular, it is shown that the analytical solutions obtained using a quadratic approximation of the energy model are very similar to the numerical solutions but they can be obtained with less computational effort, making them applicable in real-world applications.

Implicit imitation reinforcement learning (IIRL) is a framework that aims to aid a trainee agent’s learning process via observing the state transitions of a mentor, but without access to the latter's action information. Standard IIRL assumes a shared Markov decision process (MDP) between the mentor and trainee, consequently implying an identical action space. This restriction imposes limitations on the applicability of implicit imitation frameworks in real-life scenarios where, possibly due to variations in physical characteristics, the mentor agent may possess distinct own actions, thereby creating a heterogeneous action setting. In this work, we extend the deep implicit imitation Q-networks (DIIQN) method -an online, model-free, deep RL algorithm for implicit imitation- to allow for heterogeneous action sets between mentor and trainee agents. Equipped with our heterogeneous actions DIIQN (HA-DIIQN) method, a trainee agent can harvest the benefits of IIRL even in heterogeneous action settings, achieving accelerated learning and outperforming non-optimal mentor agents.

This article presents a joint trajectory optimization algorithm for a number of connected and automated vehicles in a lane-free traffic environment with vehicle nudging. A double double-integrator model is utilized for the longitudinal and lateral movements of each vehicle. The objective function consists of several sub-objectives that reflect corresponding, partially competing driving aspects and concerns, including passenger comfort, low fuel consumption, vehicle advancing at desired speed, collision avoidance, suppressing of infeasible maneuvers. Fixed and state-dependent control input bounds account for various technical limitations as well as for road boundary respect. The solution of the formulated joint Optimal Control Problem (OCP) is computed by use of a very efficient Feasible Direction Algorithm, which exploits the structure of the state equations to map the OCP into a reduced Nonlinear Programming Problem. To demonstrate the efficiency of the proposed approach, challenging scenarios are examined on a lane-free straight motorway stretch. The results of the centralized (joint) OCP are compared with a previously investigated decentralized approach where OCPs are employed separately for individual vehicles.

Implicit imitation assumes that learning agents observe only the state transitions of an agent they use as a mentor, and try to recreate them based on their own abilities and knowledge of their environment. In this paper, we put forward a deep implicit imitation Q-network (DIIQN) model, which incorporates ideas from three well-known Deep Q-Network (DQN) variants. As such, we enable a novel implicit imitation method for online, model-free deep reinforcement learning. Our thorough experimentation in the complex environment of the emerging lane-free traffic paradigm, verifies the benefits of our approach. Specifically, we show that deep implicit imitation RL dramatically accelerates the learning process when compared to a “vanilla” DQN method; and, unlike explicit imitation reinforcement learning, it is able to outperform mentor performance without resorting to additional information, such as the mentor’s actions.

Cooperative intelligent transportation systems (C-ITS) support the exchange of information between vehicles and infrastructure (V2I or I2V). This paper presents an in-vehicle C-ITS application to improve traffic efficiency around a merging section. The application balances the distribution of traffic over the available lanes of a freeway, by issuing targeted lane-changing advice to a selection of vehicles. We add to existing research by embedding multiple vehicle classes in the lane-changing advisory framework. We use a multi-class multi-lane macroscopic traffic flow model to design a feedback-feedforward control law that is based on a linear quadratic regulator (LQR). The weights of the LQR controller are fine-tuned using a response surface method. The performance of the proposed system is evaluated using a microscopic traffic simulator. The results indicate that the multi-class lane-changing advisory system is able to suppress shockwaves in traffic flow and can significantly alleviate congestion. Besides bringing substantial travel time benefits around merging sections of up to nearly 21%, the system dramatically reduces the variance of travel time losses in the system. The proposed system also seems to improve travel times for mainline and ramp vehicles by nearly 20% and 42%, respectively.

Vehicular traffic has evolved as a crucial means for the transport of persons and goods, and its importance for the economic and social life of modern society cannot be overemphasized. On the other hand, recurrent vehicular traffic congestion, which appears on a daily basis, particularly in metropolitan areas, around the globe, has been a (increasingly) serious, in fact threatening, problem that calls for drastic solutions. Traffic congestion causes excessive travel delays, substantial fuel consumption and environmental pollution, and reduced traffic safety. Conventional traffic management measures are valuable [1]-[3] but not sufficient to address the heavily congested traffic conditions, which must be addressed in a more comprehensive way that exploits gradually emerging and future ground-breaking capabilities of vehicles and the infrastructure.

Eco-driving, a timely and well-known subject, aims at reducing fuel consumption by appropriately maneuvering a vehicle with a human or automated driver. In this work, the eco-driving problem is cast in an optimal control framework. State equations reflect the simple vehicle kinematics for position and speed, with the acceleration acting as a control input. Initial and final states (position and speed) are fixed. For the fuel consumption estimation, a number of alternatives are employed. To start with, a realistic, but nonlinear and non-smooth formula from the literature is considered. Simple smoothing procedures are then applied to enable the application of powerful numerical algorithms for the efficient solution of the resulting nonlinear optimal control problem. Furthermore, simpler quadratic approximations of the nonlinear formula are also considered, which enable analytical problem solutions. A comprehensive comparison on the basis of various driving scenarios demonstrates that the often utilized, but sometimes strongly questioned, square-of-acceleration term delivers excellent approximations for fuel minimizing trajectories in the present setting. A GLOSA (Green Light Optimal Speed Advisory) approach, based on the analytical solution of an optimal control problem is also presented.

The development and deployment of simple, yet efficient, coordinated and integrated control tools for motorway traffic control remains a challenge. A generic integrated feedback-based motorway traffic flow control concept is proposed in this paper. It is based on the combination and suitable extension of control algorithms and tools proposed or deployed in other studies, such as ramp metering or variable speed limit (VSL)-enabled cascade-feedback mainstream traffic flow control, and allows for consideration of multiple bottlenecks. The new controller enables coordination of ramp metering actions at a series of on-ramps, as well as integration with VSL control actions, toward a common control goal, which is bottleneck throughput maximization. While doing this, the approach considers a pre-specified (desired) balancing of the incurred delays upstream of the employed actuators, via a suitably designed knapsack problem. Despite the multitude of the offered configurations, options, and possibilities, the generic control algorithm remains simple, efficient, and suitable for field implementation. The control algorithm is demonstrated and evaluated using a validated macroscopic traffic flow model for a number of scenarios.

This paper proposes a macroscopic model and two control algorithms for the dynamic operation of reversible lanes on freeways. The proposed model is an extension of the second-order traffic flow model METANET. The reversible lanes are modeled like variable lane drops (taking into account that the cars in the closed/opened lanes need a certain time to leave/enter the corresponding segments). Based on this model, two kinds of dynamic controllers have been developed. The first one is an easy-to-implement logic-based controller that takes into account the congestion lengths generated by the reversible lane bottleneck and uses this information for the dynamic operation of the lanes. The second one is a discrete model predictive control that minimizes the total time spent of the modeled network within some constraints for the maximum values of the generated bottleneck queues. The discrete optimization is carried out via evaluation of the cost function for all the leaves in a reduced search tree. The proposed model and control algorithms are simulated and tested using loop detector data collected over a section of the SE-30 freeway in Seville, Spain. The modeled 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 the proposed model is able to reproduce traffic congestion due to the reversible lanes and that all the proposed controllers (which can be computed in a short time) substantially reduce this congestion.

Mainstream traffic flow control (MTFC) enabled via variable speed limits (VSLs) has been investigated in previous studies, utilizing various control strategies. In this paper, an extended feedback control strategy is proposed for MTFC enabled via VSLs, considering multiple-bottleneck locations. Feedback-based results are compared with optimal control results for the evaluation of the controller using a validated macroscopic model. The performance of the feedback controller is shown to approach the optimal control results, despite the fact that many practical and safety restrictions are additionally considered by the feedback controller.

The well-known feedback ramp metering algorithm ALINEA can be applied for local ramp metering or used as a key component in a coordinated ramp metering system. ALINEA uses real-time occupancy measurements from the ramp-flow merging area that may be at most few hundred meters downstream of the metered on-ramp nose. In many practical cases, however, bottlenecks with smaller capacity than the merging area may exist further downstream for various reasons, which suggests using measurements from those further downstream bottlenecks rather than from the merging area. This paper addresses the local ramp metering problem in such a downstream bottleneck case. Theoretical analysis indicates that ALINEA may lead to a poorly damped closed-loop behavior in this case, but PI-ALINEA, which is a suitable proportional-integral (PI) extension of ALINEA, can lead to satisfactory control performance. The stability of the closed-loop ramp metering system with PI-ALINEA is rigorously proved by Lyapunov stability arguments. The root locus method is also employed to analyze the linearized closed-loop system performance of ALINEA and PI-ALINEA with and without a downstream bottleneck to provide insights on both controllers' performance. Simulation studies are conducted using a macroscopic traffic flow model to demonstrate that the ramp metering performance of ALINEA indeed deteriorates in the distant downstream bottleneck case, whereas a significant improvement is obtained using PI-ALINEA. Moreover, with its control parameters appropriately tuned, PI-ALINEA is found to be universally applicable to a range of distances between the on-ramp and downstream bottlenecks. This indicates that little fine-tuning would be necessary in field applications.

This paper presents a person-based traffic responsive signal control system for transit signal priority (TSP) on conflicting transit routes. A mixed-integer nonlinear program (MINLP) is formulated, which minimizes the total person delay at an intersection while assigning priority to the transit vehicles based on their passenger occupancy. The mathematical formulation marks an improvement to previous formulations by ensuring global optimality for undersaturated traffic conditions and intersection design and traffic characteristics that lead to convex objective functions in reasonable computation time for real-time applications. The system has been tested for a complex signalized intersection located in Athens, Greece, which is characterized by multiple bus lines traveling in conflicting directions. Testing includes cases with deterministic vehicle arrivals at the intersection and emulation-in-the-loop simulation (EILS) tests that incorporate stochasticity in the vehicle arrivals. The results show that the proposed person-based traffic responsive signal control system reduces the total person delay at the intersection and effectively provides priority to transit vehicles, even when perfect information about the auto and transit arrivals at the intersection is not available.

Recent research has proposed mainstream traffic flow control (MTFC), enabled via variable speed limits (VSLs), as a novel motorway traffic management tool and has demonstrated its efficiency based on sophisticated optimal control methods that may face difficulties in practical field implementations. A simple local MTFC feedback controller is designed in this paper, taking into account a number of practical requirements and restrictions. The MTFC controller relies only on readily available real-time measurements (no online model usage and no demand predictions are needed) and is therefore robust and suitable for field implementations. The controller is evaluated in simulation and compared with optimal control results. Despite its simplicity, the new controller's performance is shown to approach the optimal control results while considering several practical and safety restrictions for a number of investigated scenarios.

Metered dual-branch on-ramps may feature strongly different (relative) queues or waiting times on each of their branches. Different methods are developed to balance the queues or relative queues or waiting times on both branches. The methods are evaluated and compared based on extensive microscopic simulations. As a by-product, a waiting-time estimation and control method for metered ramps is also developed. The developed methods are implemented in the operational ramp-metering system of the Monash Freeway (Melbourne, Australia). The developed concepts are also applicable to other kinds of traffic control problems involving merging traffic streams.

In a recent series of articles with largely identical contents and results, some claims are raised about the pertinence and performance of the well-known and widely field-applied local ramp metering algorithm ALINEA and of some extended versions thereof. The expressed claims are based on simulation results with a self-made microscopic simulator. This paper shows that the produced simulation results and derived conclusions are based on an insufficient understanding of the feedback character of the ALINEA algorithm, which led to an inappropriate application of the method. More specifically, the mainstream measurement that feeds ALINEA was misplaced so that any occurring congestion could not be monitored; this renders ALINEA blind to the traffic conditions under control and negates the very notion of feedback.

A new traffic-responsive ramp-metering strategy is presented that coordinates local ramp-metering actions, thus enabling the linked control of the inflow from two (or more) consecutive on-ramps to the freeway mainstream. The proposed linked ramp-metering scheme is simple and utterly reactive, i.e., based on readily available real-time measurements without any need for real-time model calculations or external disturbance prediction. The well-known feedback strategy, known as Asservissement LINeaire d'Entree Autoroutiere (ALINEA), is used at a local level. Simulation results are presented for a hypothetical freeway axis with two successive on-ramps. Some pitfalls and misapplications of the local ramp metering are also illustrated via appropriately designed simulation scenarios. The proposed linked strategy is demonstrated to outperform the uncoordinated local ramp metering and, thus, to increase the achievable control benefit over the no-control case. In fact, the new strategy is shown to reach the efficiency of sophisticated proactive optimal control schemes.