Cellular Automata model for heterogeneous traffic

Journal of Advanced Transportation, Fall, 2009 by Ch. Mallikarjuna, K. Ramachandra Rao

Background

A CA approach to traffic simulation is useful since computationally efficient microscopic simulation is possible moreover it also facilitates distributed computing. This is on account of the fact that the rule set that describes the update on each vehicle is very small. Besides this; different pieces of the roadway are represented in identical or nearly--identical ways, while the update schedule is being completely parallel and is thus extremely simple. Also, a more subtle reason for the high speed of CA is because the behavioural pattern of the vehicles is computed implicitly. The benefit of using implicit computation is that it minimizes the computation and thereby increases the speed. In the CA models of traffic, the position, speed, acceleration and time are treated as discrete variables. In this approach, a lane is represented by a one-dimensional lattice and each lattice site represents a cell, which can be either empty or occupied by at the most one vehicle at a given instant of time.

Recognizing the recent developments and the advantages associated with the CA models, an attempt is made in this study to simulate heterogeneous traffic applying CA approach. From the field observations it can be seen that the traffic comprises several types of vehicles and on National Highway-l, near Delhi, it is observed that around 50% of the traffic comprised cars. Since the observed location is closer to Delhi city, the car traffic is quite high it is slightly less on the rural highways. In Table 1, the physical and mechanical characteristics of various vehicles commonly observed in heterogeneous traffic are presented (Arasan and Koshy, 2005).

Observed Heterogeneous Traffic Features

The modified CA model developed in this study assumes five types of vehicles namely car, bus, truck, two-wheeler and three-wheeler. Though the traffic consists of several other vehicles besides the five mentioned above, the present model is developed considering only these vehicles. This is due to the fact that the majority of the traffic comprises these vehicles. The main emphasis of the study is on modelling a realistic vehicle behaviour which includes maximum speed, acceleration, deceleration characteristics and vehicle dimensions. The basic structure and the updating procedure of the cellular automata model is decided on the basis of the following characteristics of heterogeneous traffic.

1. Vehicle composition

2. Vehicle width and length

3. Safe lateral spacing and longitudinal spacing maintained by vehicles

4. Acceleration and deceleration characteristics

5. Driver behaviour in heterogeneous traffic such as a no-lane discipline and overtaking from either side of the leading vehicle.

The vehicle composition is a very important parameter in deciding the basic structure of the CA model. In addition to cars, the composition of two-wheeler and three-wheeler is very significant in the traffic observed in the urban areas. However, the traffic in rural areas is observed to mainly comprise cars, trucks and buses. Furthermore, recent trends show that the proportion of cars is increasing rapidly on rural roads. Also, since there is a significant proportion of two-wheelers and three-wheelers in the traffic, now it is imperative to consider these vehicles in the model too. Physical dimension of vehicles is very crucial in deciding the cell size used in the CA model. As suggested in the original CA model, a cell of 7.5 m length may not be suitable to represent two-wheelers and three-wheelers since their length is very small.