In the field of transportation, data analysis is probably the most important and widely used research tool available. In the data analysis universe, there are two ‘schools of thought’; the first uses statistics as the tool of choice, while the second – one of the many methods from – Computational Intelligence. Although the goal of both approaches is the same, the two have kept each other at arm’s length. Researchers frequently fail to communicate and even understand each other’s work. In this paper, we discuss differences and similarities between these two approaches, we review relevant literature and attempt to provide a set of insights for selecting the appropriate approach.

Statistical methods versus neural networks in transportation research: Differences, similarities and some insights

This paper introduces Simulation of Intelligent TRAnsport Systems (SITRAS), a massive multi-agent simulation system in which driver-vehicle objects are modelled as autonomous agents. The simulation outputs can be used for the evaluation of Intelligent Transport Systems applications such as congestion and incident management, public transport priority and dynamic route guidance. The model concepts and specifications, and the first applications of the model in the area of incident modelling in urban arterial networks were described in previous publications. This paper presents the details of the lane changing and merging algorithms developed for the SITRAS model. These models incorporate procedures for ‘forced’ and ‘co-operative’ lane changing which are essential for lane changing under congested (and incident-affected) traffic conditions. The paper describes the algorithms and presents simulation examples to demonstrate the effects of the implemented models. The results indicate that only the forced and cooperative lane changing models can produce realistic flow-speed relationships during congested conditions.

Modelling lane changing and merging in microscopic traffic simulation

This paper attempts to summarise the findings of a large number of research papers concerning the application of neural networks to transportation. A brief introduction to neural networks is included, for the benefit of readers unfamiliar with the techniques. Because the subject is so young, some of the papers appear only in conference proceedings or other less formal publications. I make no apology for this; I felt it was important to cover as much of the contemporary work as was possible. The paper surveys both the application areas found to be fruitful and the range of neural network paradigms which have been used. Not surprisingly, multilayer feedforward networks such as backpropagation have so far been by far the most popular, but there are signs of a growing diversity; practitioners using neural networks are urged to seek out the less well known paradigms and experiment with them themselves. A particular weakness noted in much of the work is the informal approach taken to detailed analysis of the results of the research. It is postulated that a more rigorous approach to matters such as comparison with other techniques and also the methodology used to design the neural networks would help a clearer picture to emerge as to best practice and future research directions.

A review of neural networks applied to transport

This paper comprehensively reviews recent developments in modeling lane-changing behavior. The major lane changing models in the literature are categorized into two groups: models that aim to capture the lane changing decision-making process, and models that aim to quantify the impact of lane changing behavior on surrounding vehicles. The methodologies and important features (including their limitations) of representative models in each category are outlined and discussed. Future research needs are determined.

Recent developments and research needs in modeling lane changing

Driver behaviors, particularly lane-changing behaviors, have an important effect on the safety and throughput of the roadway-vehicle-based transportation system. Lane-changing models are a vital component of various microscopic traffic simulation tools, which are extensively used and playing an increasingly important role in Intelligent Transportation Systems studies. The authors conducted a detailed review and systematic comparison of existing microscopic lane-changing models that are related to roadway traffic simulation to provide a better understanding of respective properties, including strengths and weaknesses of the lane-changing models, and to identify potential for model improvement using existing and emerging data collection technologies. Many models have been developed in the last few decades to capture the uncertainty in lane change modeling; however, lane-changing behavior in the real world is very complex due to driver distraction (e.g., texting and cellphone or smartphone use) and environmental (e.g., pavement and lighting conditions) and geometric (e.g., horizontal and vertical curves) factors of the roadway, which have not been adequately considered in existing models. Therefore, large and detailed microscopic vehicle trajectory data sets are needed to develop new lane changing models that address these issues, and to calibrate and validate lane-changing models for representing the real world reliably. Possible measures to improve the accuracy and reliability of lane-changing models are also discussed in this paper.

Review of Microscopic Lane-Changing Models and Future Research Opportunities

Neural networks offer a potential alternative method of modelling driver behaviour within road traffic systems. This paper explores the application of neural networks to modelling the lane-changing decisions of drivers on dual carriageways. Two approaches are considered. The first, preliminary approach uses a prediction type of neural network with a single hidden layer and the back propagation learning algorithm to model the behaviour of an individual driver. A series of consecutive time-scan traffic patterns, which describe the driver's environment and changes over time as the selected vehicle travels along a link, are input to the neural network, which then predicts the new lane and position of the vehicle. Training data are collected from a human subject using an interactive driving simulation. The trained neural network successfully exhibited the rudiments of driving behaviour in terms of lane and speed changes. A major disadvantage of this approach was the difficulty in recording real-life data, which are required to train the neural network, for individual drivers. The second approach concentrates specifically on lane changing and makes use of a learning vector quantization classification type of neural network. Input to the neural network still consists primarily of time-scan traffic patterns, but the format is changed to facilitate the possibility of data acquisition using image processing. The neural network output classifies the input data by determining the new lane for the vehicle concerned. Performance in both testing and training was very good for data generated by the rule-based driver-decision model of a microscopic simulation. Performance in testing was less satisfactory for data taken directly from a road and highlighted the need for extensive data sets for successful training.

Modelling dual carriageway lane changing using neural networks

The paper considers the application of neural networks to model lane changing on dual carriageways. The first approach uses prediction type neural networks to model the behaviour of individual drivers. Neural network input principally comprises a series of consecutive time scan traffic patterns describing the driver's environment and changes over time as the selected vehicle travels along a link. The neural network then predicts the new lane and position of the vehicle. A major disadvantage of this approach is the difficulty in obtaining the data required to train the neural network. An alternative approach concentrates specifically on lane changing and makes use of classification type neural networks. Input to the neural network still consists primarily of time scan traffic patterns. The format, however, is changed to facilitate the possibility of data acquisition using image processing. The neural network output classifies the input data by determining the new lane for the vehicle concerned. Progress to date is described. (A) For the covering abstract see IRRD 861794.

J.G. Hunt

Papers

Modelling dual carriageway lane changing using neural networks

Modelling dual carriageway lane changing using neural networks

Factors Influencing the Performance of a Neural Network Driver Decision Model: A Case Study Using Simulated Data