Gaetano Fusco

Combined signal setting design and traffic assignment problem

Abstract The paper provides a comprehensive discussion about the global signal settings problem, subject to the user equilibrium constraint for traffic flows. A new algorithm that applies the Armijo rule for step size estimation to usual projected gradient algorithm is presented and compared to other usual solution procedures. Moreover, numerical experiments are performed on a test network in order to investigate the shape of the objective function and then obtain further information about mathematical properties of the problem. Issues concerning multiplicity of solutions, algorithm convergence, and sensitivity to demand patterns are also discussed.

Artificial Neural Network Models for Car Following: Experimental Analysis and Calibration Issues

This article deals with the application of artificial neural networks to model car following drivers’ behavior. The study is based on experimental data collected by several global positioning system-equipped vehicles that follow each other on urban roads. A “swarm” stochastic evolutionary algorithm has been applied in the training phase to improve convergence of the usual error-back propagation algorithm. Validation tests show that artificial neural networks (ANNs) provide a good approximation of driving patterns. Therefore, ANN can be suitably implemented in microsimulation models. In this regard, a new experimental calibration method for microsimulation software might consist of training one specific ANN for each representative individual of the driver population through systematic observations in the field or in virtual environment trials.

Heuristic methods for the optimal location of road traffic monitoring

The paper deals with two different heuristic approaches for solving the problem of optimal location of traffic count sections (OLTCS), which is a crucial issue of the design of area wide ITS traffic monitoring centers. The first method applies deterministic rules on O/D flows and O/D pairs coverage. The second is based on a genetic algorithm (GA). Both are applied on a real size extra-urban road network and compared to a state-of-the art rule widely used in practical applications

Maximal bandwidth problems: a new algorithm based on the properties of periodicity of the system

A new approach to arterial progression optimisation, based upon the properties of periodicity in time and space of the system, gives rise to the concept of equivalent systems and module of the system, which allow us to devise a very rapid algorithm for solving a bandwidth maximisation problem. Because inbound speed, outbound speed, and cycle time are synthetically expressed by the module, investigating the dependence of the solution upon these variables is greatly facilitated. The knowledge of the solution as a function of the module makes it possible to determine easily and rapidly the supremum value of the bandwidth, while the availability of a family of maximal bandwidth solutions opens new perspectives in investigating the relationship between bandwidth maximisation and delay and stop minimisation problems.

Short-term traffic predictions on large urban traffic networks: Applications of network-based machine learning models and dynamic traffic assignment models

The paper discusses the issues to face in applications of short-term traffic predictions on urban road networks and the opportunities provided by explicit and implicit models. Different specifications of Bayesian Networks and Artificial Neural Networks are applied for prediction of road link speed and are tested on a large floating car data set. Moreover, two traffic assignment models of different complexity are applied on a sub-area of the road network of Rome and validated on the same floating car data set.

A New Analytical Model for Traffic Signal Synchronization

In this paper, using some mathematical properties of the maximal bandwidth problem, an analytical formalization in close form expressing travel time and average delay as functions of maximal bandwidth variables has been found out. A calculation procedure has also been carried out and numerical examples are presented. The results exhibit a more than satisfactory efficiency of the procedure which, interestingly, can be extended to more general problems, like global network optimization.

Traffic monitoring and incident detection through VANETs

Road traffic monitoring is one of the key applications in the Intelligent Transport System field. New technologies are now provided in this field and among the most relevant ones there is the DSRC (Dedicated Short Range Communication) set of protocols and standards where vehicles wirelessly communicate. In this paper, we deal with the application of Vehicular Ad-Hoc Networks to road traffic monitoring and we present the design of two distributed protocols based on the DSRC. A realistic simulation of a main expressway in Rome, Italy, is implemented and the performances of the two proposed monitoring methodologies are evaluated in case of regular traffic conditions and in case of a car accident. In both cases the protocols are able to capture in a very quick time (few seconds) the current traffic conditions even on a quite long road of about 70 km. A discussion about the impact of the market penetration rate of the on-board DSRC devices on the protocols performance is also provided.

Last-mile, a procedure to set-up an optimized delivery scheme

This paper illustrates a comprehensive procedure for the design of a delivery scheme in an urban area, where a set of possible locations for logistic transit-points exists. In such a scheme it is assumed that deliveries for Business-to-Consumer (BtoC) e-commerce of goods are performed at specific drop-points, suitably selected to match the pick-up of the parcels to usual activities of the customers, like the breakfast at the bar or the purchase of the newspaper. The solution procedure, which integrates transportation system theory and operational research techniques, applies a disaggregate Nested Logit Model (NLM) for demand estimation, an Analytic Hierarchy Process (AHP) to compare and select possible drop-points, and a double string Genetic Algorithm (GA) to solve jointly both the problems of transit-point location and sizing and of drop-point clustering for deliveries tours. A GA performance function is computed by solving a standard Travel Salesman Problem (TSP) on the road graph, whose travel times have been estimated by assigning the O/D matrix of car trips. The first application of the procedure to the town of Terni, in Italy, has provided very encouraging results.

Solution procedures for the global optimization of signal settings and traffic assignment combined problem

In this paper, we discuss different procedures for solving the global signal settings and traffic assignment combined problem. We present a stochastic method based on a simulated annealing approach and a Projected Gradient Algorithm (PGA), which uses the Armijo rule for the step size estimation routine, and we compare them to other two usual solution procedures. Numerical experiments conducted on a test network show that the Armijo rule can improve the efficiency of PGA significantly.

O-D MATRIX ESTIMATION AND INCIDENT DETECTION IN URBAN AREAS USING ARTIFICIAL NEURAL NETWORKS

Abstract This paper discusses the application of Neural Networks to OD estimation and assignment problems. Error Back Propagation Neural Networks, when suitably constrained to respect the road network structure, show an interesting formal analogy with both these problems. It is expected that the constrained neural network, because of its own structure, would be able to reproduce feasible states of the road network that are different to those learned from the observed sample. Following this analogy, a procedure is presented to detect both the occurrence and the extent of incidents in urban road networks. The procedure is then tested, as well as assignment and OD estimation problems, in a first experiment that confirms the goodness of this approach.