Traffic patterns used to be predicted theoretically by statistical models that treated traffic as a homogeneous fluid, and sections of transport networks were analysed without regard for individual drivers or interactions between them. Traffic analysis has recently been revolutionised by advances in mathematics and computing, and its theory is somewhere between that of social systems and physical systems. Counter-intuitive effects sometimes occur, for example Braess's paradox where average traffic speed is sometimes reduced by raising a road network's capacity. A traffic system becomes unreliable and subject to breakdown as soon as it exceeds its capacity. Thus the aim is to design systems just under capacity. Complex traffic systems can sometimes show signs of self-organisation. Advances in computing are making simulations of such systems increasingly realistic. Researchers can use mathematical models and cellular automata to study the resulting aggregate vehicle flows. For example, the Transportation Analysis Simulation System (TRANSIMS) creates a laboratory model for testing traffic scenarios, which has successfully modelled observed urban traffic patterns. MIT's Intelligent Transport Systems Program imitates the behaviour of individual drivers in specific road conditions. For the covering abstract, see IRRD 896880.
Abstract