Sublane-Model

This page describes simulations with increased lateral resolution. This model is activated using the option --lateral-resolution <FLOAT>. The model is described in Simulation framework for testing ADAS in Chinese traffic situations in proceedings of SUMO2016

New Parameters

The vehicle behavior is subject to model-specific vType attributes (maxSpeedLat, minGapLat, latAlignment) and lane-changing-modell attributes (lcSublane, lcPushy).

Model Details

The regular lanes of the road network are divided into sublanes with a minimum width of the given resolution (--lateral-resolution). If the lane width is not a multiple of the given value, the leftmost sublane has a reduced with. The default lane-width of SUMO is 3.2m so a lateral resolution of 0.8 will created exactly 4 sublanes of that width per lane. A resulution of 1.0 will create three sublanes of 1.0m width and on more lane of 0.2m width. It is recommended use a resolution that is at least as small as the least wide vehicle being simulated (i.e. motorcycles).

Vehicles occupy one or more sublanes and perform car-following calculations for all vehicles that are being followed on at least one sublane. Lane-changing takes place at the sublane level and potentially uses the whole width of the road according to the width of the vehicles. Besides changing for route-following, cooperation or speed gain, vehicles also perform lateral movement to achieve a certain lateral alignment. This preference for either staying in the middle of a lane or to one of its sides is configured with the vType attribute latAlignment.

Note, that vehicles can be located laterally anywhere on the lane(s). Vehicles change lanes incrementally depending on their maxSpeedLat attribute. The width of sublanes affects the fidelity of car following in regard to the acceptance of lateral gaps and also determines the number of candidate movements that are evaluated during lane-changing.