Single vehicle network versus dispatcher: user assignment in an agent-based model

Abstract

ABSTRACTThe spread of smartphone applications, the negative social impacts of the private car or the potentially revolutionary effect of future autonomous vehicles, are contributing to the emergence of other hybrid mobility systems. Notably, companies like Uber or Lyft have successfully applied the classical dial-a-ride method, in which users are assigned to vehicles in such a way as to minimise the cost function under a set of constraints. In this paper, we construct an agent-based model based on vehicles where demand is assigned through aggregation by origin–destination pair. Two operating strategies are described: in the first, the pickup strategy is applied at vehicle level, in the second it is applied at global service level, with vehicles being shared dynamically and making extra stops to collect travellers with different destinations. At each iteration time, the set of possible options depending on itinerary and demand are tested first for already assigned vehicles then for free vehicles. The option associated with the maximum utility is selected to set the vehicle or service strategy. A case study illustrates a comparison between both strategies and evaluates the efficiency of several service choices. A service where vehicles are managed by a dispatcher is more efficient for both users and operator.