Validating a parking lot assignment method by Eclipse SUMO

Abstract

The present paper shows how a novel autonomous vehicle (AV) algorithm can be tested and evaluated by multiple simulations with different levels of refinement.
 As the first AVs will appear on our roads (possibly) within the upcoming decade, novel traffic problems, like parking lot assignment, will also emerge. Parking is that kind of activity that will fundamentally change. Today drivers attempt to find parking places close to the destinations to minimize the walking time. The autonomous vehicles will drop off passengers at their very destination, and then, on their own, will seek even a distant parking place. Such change will affect human activities, city traffic, and the parking infrastructure.
 To investigate this problem we need an abstract city model, because an attempt to create a detailed model of the city traffic from the perspective of decades may be misleading and may lead to false emergent conclusions. We report on such an abstract model and the obtained conclusions in the paper.
 A question remains how genuine is the introduced future city model. To validate it, we propose to confront it with a detailed microscopic traffic simulation on a road network borrowed from an existing city of corresponding layout and complexity. To this end, we use Eclipse SUMO simulating the traffic of an expansive Budapest district, and show that the results obtained with the simplified abstract mathematical model stay valid.