Evaluating Crowdshipping Systems with Agent-Based Simulation

Due to e-commerce growth and urbanization, delivery companies are facing a rising demand for home deliveries, which makes it increasingly challenging to provide parcel delivery that is cheap, sustainable, and on time. This challenge has motivated recent interest in crowdshipping. In crowdshipping systems, private citizens are incentivized to contribute to parcel delivery by making small detours in their daily lives. To advance crowdshipping as a new delivery paradigm, new crowdshipping concepts have to be developed, tested, and evaluated. One way to test and evaluate new crowdshipping concepts is agent-based simulation. In this paper, we present a crowdshipping simulator where the crowd workers are modeled as agents who decide autonomously whether they want to accept a delivery task. The agents’ decisions can be modeled based on shipping plans, which allow to easily implement the most common behavior assumptions found in the crowdshipping literature. We perform simulation experiments for different scenarios, which demonstrate the capabilities of our simulator.