A Modified Social Force Model for Lane-Disciplined Vehicular Traffic Simulation

This paper proposes and validates the application of a two-dimensional Social Force Model to simulate the lane-disciplined vehicular flow on roads. In detail, a lane-wise lane-keeping force field is established along the lateral direction of the road. This force, with a magnitude quadratic to the lateral displacement of the vehicle to the lane centerline, confines the lateral movement of the vehicle to form a laned pattern on the road. To realize the overtaking maneuvers and avoid collisions, the distance-dependent overtaking force and repulsive forces are also defined in the model. The parameters in the proposed model are calibrated by comparing with the fundamental diagram collected in the literature (GA400 dataset). The comparison shows that the proposed model could provide a good prediction in terms of average speed and speed range up to 80 veh/km/lane. However, for very high-density flows, the model is likely to overpredict the flow speed, possibly owing to a fixed target speed in the model. After the validation step, the model is employed to simulate traffic flows on a segment of 4-lane freeway in Nanjing, China. The comparison of the simulated and measured trajectories demonstrates that, except for very abnormal behaviors, the proposed model could predict most car-following and lane-changing behaviors based on relative velocity and inter-vehicle gaps at the microscopic level.