Modeling Capacity of Roundabouts Using Soft Computing Techniques

Abstract

The roundabout is a very common solution to the stop controlled intersections and is an increasingly common form of road junction all over the world and widely used in urban and some rural areas of India. For the effective design of roundabouts a detailed analysis of maximum vehicle throughput capacities is required termed as capacity prediction. In India the nature of traffic is heterogeneous but most of the studies related to critical gap evaluation and capacity prediction have been carried out in developed countries having homogeneous nature of traffic flow and strictly followed lane disciplines. In the study related to transportation engineering probably the most and widely used research tool accessible is suitable and proper analysis of data. In this research work video recording technique was adopted for data collection and its analysis and the gap acceptance behavior of drivers is presented for twenty seven unsignalized roundabout sites in India.Statistics and computational intelligence are the two main approaches used in this work for the purpose of data analysis. Capacity of roundabouts is mostly predicted by using regression analysis and gap acceptance-based models but their results are not satisfactory. Variation in driver behavior and prediction techniques used in various models results in difference in the predicted capacities by the various models. A new approach which is used frequently in this area is termed as soft computing. It includes various alternatives to predict the capacity of roundabouts such as artificial neural network (ANN), fuzzy logic, cellular automata, and adaptive neuro-fuzzy interface system (ANFIS). Fully empirical, gap acceptance and simulation are the three main techniques and methodology for capacity prediction models. In this study the critical gap is estimated using gap acceptance models, regression and soft computing technique such as ANN model are used for capacity prediction of roundabouts.