Extensive search for existing models of driver behaviour showed that research in this area has been restricted to modelling of driver bahaviour under car-following situations. No work was found, on models of driver behaviour under the various other driving situations. It can be said thet all the research so far has been concentrated on modelling driver behaviour in situations, where only longitudinal interactions affect the driver. Situations where, either lateral interactions alone or lateral as well as longitudinal interactions affects the behaviour have not been modelled. Consequently, only a review of existing models of driver behaviour in car following situations is presented here. Review of car-following models

The car-following behaviour is basically, a human interactive process where the driver of the vehicle attempts to reach a stable situation and maintain it by following a leading vehicle, by continuously taking correctine actions like accelerating or decelerating.

1) Pipes' theory of car-following: Pipes [2] proposed a theory of car following behaviour based on what he referred to as the "idealized law of separation",which is explained as, A good rule for following another vehicle at a safe distance is to allow yourself the length of a car for every ten miles per hour you are travelling. Such a model implies that the actions of following vehicle is only affected by the relative speed between the leadind vehicle (LV) and the following vehicle (FV).

2) Forbes' theory of car-following: Forbes' [3] modelled car following behavior by assuming that drivers choose to keep a minimum time gap from the rear end of LV. Forbes' model of car following also implies that the actions of FV is only affected by the relative speed between the LV and FV.

3) GM-model of car-following: The General motors Research Laboratories published significant amount of work on the car-following theory model in a series of papers [4,5,6].The basic idea used here is that actions of FV in terms of acceleration or deceleration is a function of single stimulus and the sensitivity of the FV to the stimulus under the prevailing conditions.The stimulus is assumed to be the relative speed between LV and FV. Sensitivity to the stimulus is assumed to be affected by the distance headway between LV and FV, and the speed of FV.

4) Other models: eg: Rockwell_et_al [10] presents a regression based on car-following model, and Chakroborty and Kikuchi [7] a Fuzzy-Inference-based-car-following.

In trying to develop a model for driver behaviour we think it will be suitable to repersent the different features of a road environment as obstacles, which a driver needs to avoid in order to reach his destination safeley. Further, the driver would like to avoid these obstacles in such a manner, so that his speed is not impeded to a large extent.A subsidiary task is to develop a procedure which will suggest a feasible position on road for the driver.

The theory of robot motion planning based on potential field approach comes in extremely useful while trying to develop a model of driver behaviour on the lines described above. In this approach, a robot's path in an area strewn with obstacles is determined. The features of determination of such a path, which are relevant to the proposed principle of modelling driver behaviour are as follows:

Each obstacle is assumed to emit a repulsive force around it.This repulsive force is termed as a potential field and can be thought as a means of qualifying the threat posed by the obstacle to the movement of robot.
Any point in the area, where the robot can possibly move is assumed to possess a potential.This potential is the sum total of the potentials at that point due to all the obstacles in the area.
The robot selects the next point to move to by determining which of the points has the least potential.
The velocity of the robot depends on the potential of the position occupied by it. Higher is the potential, nearer the object is to the goal and hence lower should be its speed.
The Path of the robot is obtained by joining such paths.

LITERATURE REVIEW