Flow of autonomous traffic on a single multi-lane street

We investigate the behaviour of an original traffic model. The model considers a single multi-lane street, populated by autonomous vehicles directed from either end to the other. Lanes have no intrinsic directionality, and the vehicles are inserted at random at either end and any lane. Collision avoidance is fully automatic and, to enhance the transport capacity of the street, vehicles form trains in which they may travel at high speed quite close to the vehicle in front. We report on the transit times for vehicles under a wide variety of conditions: vehicle insertion probability & imbalance and their maximum speed distribution. We also outline an interesting feature of the model, that the complex interactions of many vehicles are considerably more powerful than a simple ”keep left” directive which each vehicle should obey.