Modeling city logistics using adaptive dynamic programming based multi-agent simulation

The effects of city logistics solutions are uncertain due to fluctuating demand, parking issues and multiple agents within the system. This research modelled the behavior of freight carriers and an Urban Consolidation Center (UCC) operator using Multi-Agent Simulation-Adaptive Dynamic Programming based Reinforcement Learning (MAS-ADP based RL) to evaluate a Joint Delivery Systems in an uncertain environment. The MAS-ADP based RL is superior to MAS-Q-learning in replicating the potential actions of the agents under uncertain environment by adapting to the changing environment properly into accurate decisions thus increasing the accuracy of agent's decision making and eventually reducing environmental emissions as well.