Martin Gregurić

Cooperative ramp metering simulation

The increase of vehicle numbers in recent decades resulted in road traffic congestion problems. Such congestions are a characteristic of densely populated urban areas and occur daily during morning and afternoon rush hours. Urban areas have been suffering from the lack of space needed to build new road infrastructure. The traffic congestion problem can be solved by applying new traffic control approaches from the domain of intelligent transportation systems (ITS). One of the applied methods from ITS is known as ramp metering and is used to increase the throughput of urban highways with many on- and off-ramps. Nowadays ramp metering is used in cooperation with additional control approaches like variable speed limit control (VSLC). Prior to implementation, such cooperative traffic control systems have to be tested in simulations using real world traffic data. One of the used simulators is CTMSIM which enables macroscopic simulation of highway traffic and local ramp metering approaches. In this paper the CTMSIM simulator is augmented to enable simulation of cooperative ramp metering algorithms, stand-alone VSLC, and cooperation between ramp metering and VSLC. Augmented simulator is tested using some limited available traffic data with the Zagreb bypass urban highway as a case study.

A Neuro-fuzzy Based Approach to Cooperative Ramp Metering

To solve todays road traffic congestion problems new solutions in the form of advanced control approaches of existing road infrastructure. Such solutions are from the domain of intelligent transportation systems and include various services. Technologies such as advanced driver assistant systems, and communication between vehicles and the road infrastructure are enabling new possibilities in traffic control. Vehicles can obtain a control input from the traffic management system and become an actuator ensuring that the driver complies to the traffic control system. In this paper, a concept of possible automatic vehicle control in cooperation with neuro-fuzzy based urban highway control systems (ramp metering and variable speed limit control) is described. Implemented urban highway control systems are tested using the CTMSIM simulator assuming that all vehicles support automatic vehicle control.

Ramp Metering on Urban Motorways

Reduced Level of Service on urban motorways, which actually represents evolved urban bypasses, is the product of two overlapping problems. First of them is related to the heavy congestions, and the second one is related to the unavailable constructional build-up of their capacities since they are surrounded by the urban and traffic infrastructure. In order to cope with those problems, it is necessary to introduce urban motorway control methods such as for example ramp metering (RM). The main goal of RM is to increase the throughput of urban motorways by restricting access of on-ramp traffic to mainstream traffic by using special traffic lights. In this paper, an overview of currently most used RM algorithms (ALINEA, SWARM, and HELPER) and their fundamental deficiency in partial problem solving for different traffic scenarios is given. A new RM algorithm based on the Adaptive Neuro-Fuzzy System neural network called INTEGRA is also described.

Comparison of Two Controllers for Variable Speed Limit Control

One of the traffic control methods from the domain of intelligent transport systems used to reduce congestions on urban motorways is variable speed limit control (VSLC). Main goal of VSLC is to reduce the traffic flow speed and to homogenize the traffic flow. Results are reduced density, increased safety, reduced vehicle emissions and improved throughput. To choose the best controller for VSLC prior testing in simulations is necessary. In this paper a simulation framework that enables testing and comparison of controllers for VSLC regarding traffic and environmental parameters is implemented. Using the imple-mented simulation framework two simple reactive controllers for VSLC are compared using a realistic urban motorway model.

Learning-Based Control Algorithm for Ramp Metering

Significant slowdowns in road traffic induced by increased traffic demand cause breakdowns and, consequently, congestion on roads. On urban highways, these congestion problems are most noticeable near on-ramps. To resolve traffic congestion on urban highways, it is necessary to apply new traffic control approaches like ramp metering, variable speed limit control (VSLC), etc. Today’s cooperative ramp metering algorithms adjust the metering rate for every on-ramp according to the overall traffic state on the highway and can establish additional cooperation with other traffic control subsystems. To avoid some problems of usability and effectiveness of today’s complex highway control systems, an approach based on autonomic properties (self-learning, self-adaptation, etc.) is proposed in this chapter. A new cooperative control method based on an adaptive neuro-fuzzy inference system is described. It can establish cooperation between VSLC and ramp metering. The new solution is tested using the CTMSIM macroscopic highway traffic simulator and Zagreb bypass as test model.

Ramp Metering Control Based on the Q-Learning Algorithm

Abstract Modern urban highways are under the influence of increased traffic demand and cannot fulfill the desired level of service anymore. In most of the cases there is no space available for any infrastructure building. Solutions from the domain of intelligent transport systems are used, such as ramp metering. To cope with the significant daily changes of the traffic demand, various approaches with autonomic properties like self-learning are applied for ramp metering. One of these approaches is reinforced learning. In this paper the Q-Learning algorithm is applied to learn the local ramp metering control law in a simulation environment, implemented in a VISSIM microscopic simulator. The approach proposed is tested in simulations with emphasis on the mainstream speed and travel time, using a typical on-ramp configuration.