An Order Control Policy in Crowdsourced Parcel Pickup and Delivery Service

Crowdsourced parcel delivery service has progressed dramatically by actively incorporating innovative technologies and ideas. Yet, maximizing profitability of this new type of delivery service becomes another challenge for service providers as market grows. In this paper we study a service order control policy to maximize profitability from a service provider perspective. Specifically, we suggest an order admission control approach that determines acceptance or rejection of an incoming order according to its profitability characteristics. For this, we model the problem as an average reward Semi-Markov Decision Process and utilize reinforcement learning to obtain an optimal order control policy that maximizes overall profitability of a service provider. Through numerical illustrations, we show that our suggested approach outperforms traditional methods, especially when the order arrival rate is high. Thus, smart order management is an important component of parcel pickup and delivery services.