The purpose of this research is to help companies identify in what geographical settings autonomous vehicle assisted delivery is likely to be most valuable in improving last-mile delivery and thus where they should direct their investments. The research question is motivated by both our own previous work as well as the actions of delivery companies. To this end, we model the Capacitated Autonomous Vehicle Assisted Delivery Problem (CAVADP) on a general graph. The CAVADP is the problem of using an autonomous vehicle to both drop off and pick up a delivery person at selected locations to omit the need for parking. We consider a general graph to model real-world geographies across urban to rural settings. To answer the questions about the value of autonomous vehicle assisted delivery in different geographies, we identify a number of properties of the optimal solution that allow us to reduce the problem size and introduce valid inequalities. We conduct a case study by generating test instances that reflect real-world geographies. We present extensive computational results that show the savings from the use of an autonomous vehicle is significant in all customer geographies. On average, a delivery person saves more time in urban environments than in rural environments.
Back to Workshop III: Large Scale Autonomy: Connectivity and Mobility Networks